Watch what Pakistan has done, not what it has said, because the two are pulling in opposite directions and the gap between them is the relay that the Pakistani state has been participating in.

What it has said is loud. On 9 September, Defence Minister Khawaja Asif warned that Houthi attacks on Saudi Arabia could activate the Makkah pact. Turkish media reported Islamabad was weighing airstrikes against the Houthis. The ISPR said Pakistan “stands with Saudi Arabia.” Sharif condemned the attacks. Saudi Arabia formally requested military support. The words are the words of an ally about to fight.

What it has done is nothing. On 10 September, one day after Asif’s warning, Pakistan said there were no talks on military action, and that it would act “when the time comes.” That phrase is worth bearing in mind because it is not a commitment. It is the sound a state makes when it wants credit for a promise it does not intend to keep. The time, one suspects, is always coming and never quite here.

Understand the position Pakistan is actually in, and the hedge stops looking like weakness and starts looking like the most disciplined play on the board. Islamabad shares a long border with Iran. It has a large Shia population it cannot afford to inflame. It fought the Iran war of this year not as a combatant but as a mediator, the one actor who hosted the ceasefire talks and came out with its standing improved. And it signed, first in September 2025 and then in expanded form at Mecca in August 2026, a mutual-defence agreement that treats an attack on Saudi Arabia as an attack on itself. Pakistan has promised to defend a kingdom against an axis that operates on Pakistan’s own doorstep, and it has no intention of doing the defending if it can possibly avoid it.

The guarantee is an asset. Firing it would be spending the asset. Pakistan would rather hold it forever.

This is not hypocrisy. It is arbitrage. Pakistan has worked out that the value of a defence guarantee lies almost entirely in its being believed, and hardly at all in its being used. So it collects the returns of the promise, the Saudi money, the strategic prestige of being the Muslim world’s nuclear guarantor, the 13,000 troops stationed in the Kingdom since the spring drawing Saudi gratitude and Saudi funds, while paying none of the costs, which would be a war against the Houthis that it cannot win, cannot end, and cannot survive politically at home.

The nuclear point

The nuclear question is where this reaches its finest point, and it is worth being precise, because the ambiguity is not accidental. Asif said Pakistan’s capabilities “will be made available” to Saudi Arabia under the pact. Days later he said nuclear weapons were “not on the radar.” Read together, the two statements cancel out, and the cancellation is the product. Chatham House called it a precedent for extended deterrence by a nuclear-armed state outside the treaty. But an explicit umbrella would be a catastrophe for Pakistan. It would force Islamabad to own, in writing, a commitment to risk nuclear war for Riyadh, and it would set off the proliferation cascade across the Gulf that everyone has feared for a generation. So Pakistan says the umbrella exists and does not exist in the same breath, and Iran is left to guess, and the guessing is worth more than any signed clause, because it deters without ever obliging Pakistan to act.

Reading the map

The whole structure works only as long as no one forces the question. That is the tightrope. Saudi Arabia built this alliance to deter exactly the attacks it is now absorbing, and the attacks arrived before the deterrent could set, which means the pact is being tested in its infancy, before anyone wants to find out what it is worth. The Israeli press has already called the strikes near Mecca an exposure of the pact’s limits. If the Houthis land a blow catastrophic enough, a mass-casualty strike on a Saudi city, a hit on Mecca itself, Pakistan’s ambiguity collapses into a choice, and every choice is bad. Fight, and it is in a war it dreaded. Refuse, and the guarantee it has traded on is revealed as paper, and the asset is worth nothing the moment it is seen not to work.

So the war, from Pakistan’s vantage, unfolds along the one path Islamabad will do everything to hold open: the middle. Condemn the attacks, stand in solidarity, station the troops, invoke the pact rhetorically, and never once send a jet against the Houthis. Deter by ambiguity. Profit by restraint. Keep the promise unspent. It is the cleverest position in the region and also the most fragile, because it depends on an adversary choosing never to call the bluff, and adversaries, in the end, are not in the business of leaving bluffs uncalled.

Pakistan has spent a year turning proximity to a war into a source of profit, first as the mediator who ended the fighting, now as the guarantor who need not fight. It is a remarkable trick. The question is not whether it is clever. It is how long a state can be paid for a promise it is arranging, with great skill, never to honour, and what happens on the morning someone finally presents the bill.

There is a Dovlatov story we keep coming back to, it ran in the New Yorker in 1986 and it is called “The Colonel Says I Love You”, and the plot, if you can call it a plot, is that a man wakes up after a party and there is a woman asleep on his couch. Lena. She has breakfast, she says she will be back around six, she comes back that evening and she never leaves. That’s the courtship. They marry, they have a daughter, and through all of it she is completely calm and completely unreadable and he never quite works out why she is there. Nobody says I love you. Not once, as far as we can tell.

Then Lena decides to emigrate, takes the daughter, goes. He stays, because he is a writer and he is stubborn and the Soviet Union won’t publish him anyway. And here is the bit that gives the story its title. He gets called in, and a KGB colonel at the immigration office tells him, very politely, that he should think about leaving too. He says but we’re divorced. And the colonel says no no, divorce would be a mistake, we would like to see your family reunited.

So the secret policeman is the one who makes the speech about the marriage. The only person in the whole story who argues for love is the man from the KGB, and he is arguing for it because he wants a nuisance out of the country.

(Dovlatov, by the way, had a rule that no two words in a sentence could start with the same letter. Imagine doing that in Russian, or in anything. Brodsky said he was the only Russian writer people would read all the way through, which is probably unfair to a few others but you see what he meant.)

Anyway. We thought of the colonel this week because China has put Jane Eyre on the mandatory reading list for 14-year-olds, and Beijing Daily has explained that this is so children learn about real love and not the looks-and-money version they get from social media, right at the age when they start having feelings. Which is sweet, sort of. And it is also the colonel. Because why does the party suddenly care about true love? Marriages fell by a fifth in 2024, lowest on record, and no marriages means no babies, and so here comes the state with a Victorian novel saying we would like to see your family formed.

What are they up against though? Micro-dramas, those vertical phone serials, a few minutes an episode, which went from under a billion yuan in 2020 to a hundred billion in 2025. And it’s all flaunted wealth and wronged lovers and brooding billionaires and mistresses plotting revenge. The regulator went after them in June, a list of eight banned things, wealth-flaunting and revenge and distorted views of marriage among them. But you can’t really ban taste, can you, you ban the billionaire CEO and you get the billionaire general, people want what they want.

So the textbook. The one place where nobody can swipe away.

And this is the thing we can’t get over. Have they read Jane Eyre? Poor plain orphan, rich brooding employer, secret wife in the attic, wedding stopped at the altar, she runs, she inherits money, she comes back and the house has burned down and he’s blind and sorry. That is a micro-drama. That is every micro-drama. They banned the genre and then assigned the original.

The difference I think is that Brontë gives you five hundred pages inside Jane’s head and the whole point is she says no when the terms are wrong. The ninety-second version keeps the mansion and the fire and loses the no. Maybe that’s what Beijing is after. Maybe.

But books don’t do what they’re told. A mother wrote on Xiaohongshu that her son couldn’t understand why Jane marries a blind man, he thought it was a bad deal, and honestly that boy has already been educated, just not by the school. And a sociologist at Tsinghua is worried about the opposite, that kids will take in the madwoman in the attic and all the colonial stuff without questioning any of it. And then there are the girls. Chinese readers have always read Jane as the one who won’t be pushed around, and she’s arriving in classrooms right when feminist talk is getting squeezed. “I am no bird; and no net ensnares me.” You set that for an exam and you don’t get to choose which line they underline.

The colonel did get his way, Dovlatov left, ended up in Queens, the family was reunited. At the end of the story Lena calls from abroad and says we’ll see each other if you love us. And he says what’s love got to do with it, love is for the young, this is fate. Which is not something a colonel can issue. Or a textbook, probably. We’ll see.

The New Delhi summit tested a problem created by BRICS’s own expansion. The grouping now includes eleven states with different security relationships, competing regional interests and highly unequal economic weight. Iran and the UAE disagree over the regional security order; India and China continue to manage a disputed border alongside a large trade imbalance; several members maintain close security or economic relationships with the United States.

The relevant measure of the summit, therefore, is not whether BRICS acted as a unified geopolitical bloc. It is whether the expanded grouping produced agreements in areas where its members’ interests overlap despite these differences. Four developments in Delhi help answer that question.

Key Announcements from the Joint Statement

The New Delhi Declaration retained familiar BRICS positions on reforming the UN and international financial institutions, increasing developing-country representation and opposing unilateral tariffs and coercive economic measures. More useful for assessing the grouping are the mechanisms members agreed on.

Cross-border payments were one. BRICS did not announce a common currency. Instead, the declaration continued work on interoperability between national payment and financial messaging systems, greater settlement of trade and investment in local currencies, and increased local-currency financing through the New Development Bank. It also supported phased development of a New Investment Platform. The BRICS Payment Task Force acknowledged that there is “no one-size-fits-all approach”.

This places an important limit on claims about de-dollarisation. BRICS is not constructing a monetary union. Its members are developing additional channels through which some transactions can occur without requiring dollar settlement. These mechanisms can lower transaction costs and, for countries vulnerable to sanctions, reduce exposure to disruptions in existing payment systems. Their current scale, however, is nowhere near enough to replace the dollar-centred financial system.

The declaration therefore points towards redundancy rather than replacement: maintaining access to existing institutions while developing additional mechanisms where dependence has become a source of economic risk.

UAE – Iran Meeting

The meeting between Iranian President Masoud Pezeshkian and Abu Dhabi Crown Prince Sheikh Khaled bin Mohamed bin Zayed was the highest-level UAE-Iran contact since the 2026 war. The discussion covered bilateral relations, de-escalation and regional stability. No settlement of their security disagreements or new regional arrangement was announced.

The meeting’s significance is clearer when compared with what happened earlier in the year. BRICS foreign ministers failed to produce a joint statement in May partly because Iran and the UAE could not agree on language concerning West Asia. In Delhi, the eleven members did agree. Reporting by the Indian Express indicates that reaching the compromise required extensive negotiations involving India, Iran, the UAE, as well as China, Russia, Brazil and Egypt. The final language did not resolve the underlying dispute. It emphasised sovereignty, civilian protection, restraint and continued energy and trade flows in terms broad enough for both countries to accept.

It is a limited but identifiable institutional outcome. Between May and September, BRICS moved from being unable to produce common language on the conflict to containing the disagreement sufficiently to adopt a declaration and host bilateral contact between the two sides. Whether that contact leads to sustained de-escalation remains unknown. Delhi demonstrated that membership can provide an additional channel for managing a dispute, even when BRICS lacks the authority or common position required to resolve it.

Structural Dialogues across Poly-ruled states

The economic discussions expose a different limitation. Greater trade between BRICS countries is frequently presented as evidence of growing South-South economic integration. India’s experience shows why the composition of that trade matters as much as its volume.

The clearest case is China. India’s bilateral trade deficit reached approximately $112.16 billion in FY26. Indian exports remain concentrated in commodities and relatively low-value products, while imports from China include machinery, electronics, integrated circuits, batteries, fertilisers and industrial inputs. India consequently depends on Chinese goods in sectors that are themselves important to expanding domestic manufacturing.

The Modi-Xi meeting in Delhi explicitly addressed this problem. The two sides agreed to discuss the structural trade imbalance, supply-chain concerns and more predictable market access for Indian businesses. This followed a gradual stabilisation of bilateral relations after the 2020 border confrontation, including resumed flights, visa easing and renewed diplomatic engagement, while the boundary dispute itself remains unresolved.

This distinction matters for evaluating BRICS economic cooperation. Moving trade away from Western markets does not necessarily reduce economic dependence if it increases concentration elsewhere. China’s manufacturing capacity gives it a structural advantage within BRICS that institutional declarations cannot remove.

For India, the question is consequently not simply how to increase intra-BRICS trade, but how to alter its composition. Market access, domestic manufacturing capacity and supply-chain diversification matter more to reducing the deficit than political agreement at BRICS. Diplomacy can create the negotiating channel; it cannot by itself correct the productive asymmetry that generates the deficit.

Possible alternative to American Shield?

BRICS has made some progress where alternatives can be created transaction by transaction. Local-currency settlement can reduce the need for dollars in particular trades. New Development Bank lending provides another source of development finance. Payment-system interoperability could eventually provide additional routes for cross-border transactions. None requires members to agree on a common adversary or foreign policy.

The American role in the Gulf rests on military capabilities and long-standing bilateral arrangements: bases, intelligence, logistics, air and naval forces, missile defence and the capacity to reinforce partners during crises. BRICS has no equivalent collective capability. It has no mutual-defence commitment, integrated military command or common definition of the threats against which members would be protected. Its membership makes such convergence particularly difficult. Iran seeks a reduced American military presence in the Gulf. The UAE still depends substantially on American security cooperation. India works closely with the United States while maintaining its relationship with Russia and managing competition with China. India and China themselves have not resolved their border dispute.

BRICS can therefore supplement existing arrangements more readily than it can replace them. Payments, development finance and diplomatic consultation allow selective cooperation because they do not require members to identify the same security threats. A collective security system would.

Delhi consequently provides a more limited measure of BRICS’s development. Its members can increasingly separate areas in which cooperation is possible from disputes they cannot resolve. That worked sufficiently to produce a declaration despite Iran-UAE differences, to keep India-China economic negotiations open despite continuing strategic competition, and to advance payment cooperation without monetary integration.

It does not yet amount to an alternative security order. For now, BRICS’s institutional capacity lies in managing specific disagreements and creating supplementary economic mechanisms, rather than replacing the political, financial and security structures within which its members continue to operate.

A useful test of any government is what it does when it isn’t performing for an audience. The last two weeks of August and the first of September gave China’s local wire services occasion to report both ends of that test at once: a punishing run of climate disasters across five provinces, and a quieter, province-by-province effort to keep household welfare intact underneath them. The national desk tends to smooth the two into separate stories. Read side by side, they describe the same governing instinct.

On 26 August, a mudslide struck the border port at Gyirong County, in the Xizang Autonomous Region, on the crossing with Nepal. Early reporting from Global Times and Xinhua put the toll at three dead and 265 missing. The political response moved faster than the recovery: Xi Jinping is reported to have chaired a Party meeting on relief within the disaster’s first days, a level of direct involvement that signals the event cleared whatever internal threshold triggers top-tier attention. Notably, the same reporting places the first rescue teams reaching the core disaster area only on the Friday afternoon following the slide, a gap between political mobilization and physical access that is worth sitting with, since it suggests the terrain itself, not bureaucratic delay, set the pace of the response. One detail is worth flagging for anyone tracking this story across outlets: Global Times describes the activated response as Level I, China’s highest tier, where the ministries’ own announcements carried in Xinhua’s English service elsewhere describe a Level II activation. The two are not reconcilable from the material at hand, and the discrepancy is itself a small data point about how disaster classifications get reported differently depending on which desk is doing the reporting.

Gyirong was not an isolated shock. The same window produced a landslide in Pengshui County, Chongqing, that killed 11 and left 50 missing; Typhoon Narra, the season’s nineteenth named storm, making multiple landfalls in Guangdong; a flash flood in Weiyuan County, Gansu; and a forest fire on Yuping Mountain in Hongya County, Sichuan, that was brought under control without casualties. Taken as a set tracked by ecns.cn’s society desk and the Global Times society index, this is not a cluster of unrelated weather events so much as a season in which nearly every major disaster-response category in the state’s toolkit was exercised within a matter of weeks: landslide, typhoon, flash flood, wildfire. A system tested this many times, this close together, either reveals its weaknesses or demonstrates that the weaknesses have already been priced in.

Running underneath the disaster coverage, and almost never sharing a headline with it, was a second track of governance concerned with keeping households solvent. Qinghai’s civil affairs department distributed 1.391 billion yuan, roughly $193.7 million, in relief and subsidy funds during the first half of the year. Nationally, tax and fee cuts tied to elderly care, health, education, and employment rose 11.8 percent year on year over the same period, according to State Taxation Administration data. These are not disaster-response figures; they describe an entirely separate administrative rhythm, one that accrues in aggregate over a fiscal half-year rather than announcing itself in a press conference. That the same government agencies capable of activating four-tiered emergency responses within hours are also running slow, cumulative subsidy programs at provincial scale is not a contradiction. It is closer to a division of labor between the acute and the chronic.

China Daily’s columnists made an argument this week that belongs in the same frame: that the country’s life-expectancy targets are, in practical terms, a rural problem before they are anything else, and that the community canteens now spreading through Chinese cities function as a kind of intergenerational welfare infrastructure, feeding both elderly residents who no longer wish to cook and workers who cannot afford to eat out. Whether or not one accepts the columnists’ framing, the underlying claim is consistent with the Qinghai and tax-cut figures: welfare provision in China is increasingly delivered through mechanisms, canteens, tax relief, direct subsidies, that don’t require a crisis to justify their existence, and that operate on a timeline measured in years rather than news cycles.

Not every story this fortnight arrived through official channels. Drone footage published by SCMP on 7 September showed barges discharging an unidentified black substance into the Yangtze near Tianxingzhou islet, in Wuhan, a detail that reached the public because someone happened to be filming, not because an agency disclosed it. It is a minor story set against a mudslide with hundreds missing, but it is the one item in the week’s record that arrived through observation rather than announcement, which makes it worth noting on its own methodological terms.

The week’s other register was more familiar: the arrow of construction that Chinese state media reaches for regardless of what else is happening. The Hangzhou Bay Cross-Sea Railway Bridge topped out its three main towers, and a trial train ran on the new Hangzhou–Deqing intercity line on 1 September. The National Development and Reform Commission, meanwhile, issued a seawater desalination action plan targeting 4.5 million tonnes of daily capacity by 2030. That a decade-scale infrastructure target was announced during an active, high-casualty rescue operation says something about how insulated the planning apparatus is from the emergency apiuparatus: neither one appears to disrupt the other’s calendar.

Set next to each other, the mudslide at Gyirong and the subsidy flow through Qinghai’s civil affairs department measure the same capacity from opposite directions. One is the state’s ability to move thousands of personnel and equipment into hazardous terrain within hours of catastrophic failure. The other is its ability to sustain, quarter after quarter, the transfers that keep households solvent when nothing has gone wrong at all. Neither is more revealing than the other on its own. Together, they describe a government organized around absorbing shocks geological, meteorological, or economic, before they reach the level of the household, and doing so whether or not anyone outside the local wire is paying attention.

South Asia does not need to be convinced that climate change is dangerous. It needs to recognise what kind of danger it has become. The climate-security debate is often framed too narrowly. The question is not whether a flood will cause a war or whether drought will automatically produce violence. Climate change rarely works through such direct chains of causation. Its security significance lies elsewhere: it is changing the physical conditions in which states defend territory, exercise authority, and maintain the systems on which national power depends.

That distinction matters particularly in South Asia, where some of the region’s most strategically sensitive spaces are also among its most climate-vulnerable. India and Pakistan remain locked in a nuclear rivalry. India and China contest a heavily militarised Himalayan frontier. The Himalayas contain not only glaciers and river systems but military deployments, roads, tunnels, hydropower projects and critical border infrastructure.

Climate change is not replacing these security challenges. It is altering the environment in which they have to be managed. The Himalayas make this especially clear. According to ICIMOD’s 2026 assessment, glacier ice loss across the Hindu Kush Himalaya has doubled since 2000. Rapid cryospheric change is making mountain terrain and river systems less predictable while increasing the risks of floods, landslides and glacial hazards. In a heavily militarised and strategically contested region, this is not simply an environmental problem. It threatens the infrastructure through which states maintain territorial presence and connectivity.

The recent disaster in Nepal illustrates the point. A massive mountain collapse and flood devastated infrastructure in the Nepal-China border region, demonstrating how cascading Himalayan hazards can destroy roads, bridges and energy infrastructure within hours. The significance of the event was not that it triggered military conflict. It was that a climate-related hazard could abruptly disrupt movement, connectivity and state capacity in a sensitive borderland.

Climate change also threatens the economic and resource foundations on which security rests. Extreme heat is already reducing people’s capacity to work across agriculture, construction, and the informal economy. A recent World Bank assessment estimates that extreme heat is costing South Asia the equivalent of nearly 31 million full-time jobs annually and could reduce the region’s economy by almost 7 per cent by 2050.

Those losses matter beyond individual livelihoods. States cannot indefinitely sustain strategic autonomy, military capability or political stability if climate shocks steadily weaken economic growth and strain public finances. Climate risk enters national security long before it reaches a border or requires military deployment.

Water adds another layer of pressure. The Hindu Kush Himalaya feeds the Indus, Ganges and Brahmaputra systems that underpin food, energy and livelihoods across the region. Glacier loss and increasingly erratic rainfall are making those systems harder to predict. Climate change will not automatically turn water disputes into wars, but it raises the stakes of managing shared resources in a region already marked by mistrust.

This is South Asia’s central climate-security contradiction: its climate systems are regional, but its security thinking remains overwhelmingly national and adversarial. Rivers and hazards cross borders. The institutions required to manage them often do not. Yet South Asia’s security establishments remain primarily organised around protecting states from hostile neighbours rather than managing risks that demand cooperation with them.

Environment ministries can formulate adaptation plans, and disaster agencies can respond when catastrophe strikes. But neither controls the strategic resources required to protect critical infrastructure, incorporate climate vulnerability into national planning or manage transboundary risks. Those decisions sit with finance ministries, foreign-policy establishments, defence bureaucracies and national security institutions.

A climate-security constituency would not mean militarising climate policy. Nor would it mean treating climate migrants as threats. It would mean recognising that climate risk now affects the material foundations of conventional security and must therefore enter the institutions that make strategic decisions. A state can have a powerful military and still become less secure if its border infrastructure repeatedly collapses, its energy systems become unreliable, its workforce becomes less productive, and its civilian institutions are overwhelmed by recurring disasters.

South Asia’s security establishments have spent decades preparing for threats that cross borders with hostile intent. The harder challenge is preparing for one that crosses borders without intent at all, but can still reshape a state’s ability to defend territory and maintain critical systems.

The People’s Bank of China published its five-year plan on the tenth of August. Five task areas, nine action plans attached. Most of it reads the way these documents always read. One small thing in it is worth stopping on, and it is not a phrase. It is an order.

The second task area is about serving the real economy. Inside it, the bank lists its instruments in a line. Technology finance. A bond-market technology board. Green finance. Inclusive finance. Pension finance. Digital finance. Then, at the end, boosting consumption.

Consumption is last. That is the piece worth writing about.

China does not have a supply problem. It makes more than the world can absorb. It has a demand problem. The apartments do not sell. The savings sit in the bank. Retail spending crawls. Households expect their own incomes to fall, and they are spending like people who expect that. Every serious read of the economy lands in the same place. The engine is demand, and demand is broken.

A plan built to fix that would put consumption first. This one puts it sixth, one clause after pension finance. Not by accident. The bank knows how to build the six things ahead of consumption. It does not know how to build consumption.

Look at what those six things share. A technology-finance system is something you construct. You set up the lending vehicles, name the eligible sectors, direct the credit, and it exists. A bond-market technology board is something you construct. Write the listing rules, open the board, done. Green finance, pension finance, digital finance (the whole list above the last line) are architecture. The state designs them and commands them into being. China is very good at this. Better, probably, than any state has been.

Consumption is not architecture. There is no vehicle you can set up that makes a man spend. No board you can open that gives him back his confidence in next year. You cannot point credit at his fear the way you point it at a chip plant. Consumption is a mood held by a billion people at once. You do not build it. You earn it, slowly, by giving people reasons to think the years ahead beat the years behind. That belief is what the property boom ran on and the property bust burned down. No action plan brings it back, because it was never the kind of thing an action plan makes.

So the order is not a ranking of what matters. It is an admission of what the bank can do. The things it knows how to build come first. The one thing it cannot build comes last. A man reaches first for what is in his hand. Technology finance is in the bank’s hand. The Chinese consumer’s confidence is not.

This is the part the West gets backwards about Beijing. The usual complaint is that the state controls too much, plans what should be left alone. The real trouble runs the other way. Some problems planning cannot reach. Demand is one of them. And the Chinese state has spent so long winning at the problems planning can reach that it now brings the same method to a problem built to defeat it. It is treating a crisis of confidence with a construction manual. It funds the supply of things into an economy short of the will to buy them, because supply is what it knows how to fund.

The whole plan holds up under this one lens. The technology board, the innovation finance, the digital-yuan rails, the opening of the markets. All of it the state building, straight and fast, at what it is good at. And underneath it, the saver in Chengdu who will not spend. He appears in the document once, at the end of a list, because the people who wrote it do not know what to say to him. They have built him a superb financial system. They cannot make him believe in it.

A five-year plan that can construct everything except the thing the construction is for is not a plan for recovery. It is a very good way of doing what you already know how to do, and hoping the rest sorts itself out. It will not sort itself out. That is the part no list can reorder.

As ten heads of government gather in Bishkek for the Shanghai Cooperation Organisation’s twenty-fifth anniversary, the organisation looks more consequential than it has ever been. China, Russia, India, Pakistan, Iran and the four Central Asian republics now sit inside the same institution, alongside Belarus. Together, its members account for more than three billion people and roughly a quarter of global GDP. Yet size has not made the SCO particularly capable of resolving the disputes between them. India and Pakistan remain adversaries; India and China still manage a contested border; Russia and China cooperate closely without sharing identical interests in Central Asia. The SCO is good for something more limited: it keeps security bureaucracies cooperating, gives Central Asian states a forum containing both Russia and China, and allows governments that do not trust one another to remain in regular contact without requiring political alignment.

That sounds modest because the organisation began with a modest problem. The SCO’s predecessor, the Shanghai Five, emerged after the Soviet collapse left China facing newly independent states along what had been a heavily militarised Sino-Soviet frontier. China, Russia, Kazakhstan, Kyrgyzstan and Tajikistan signed agreements in 1996 and 1997 to build military confidence and reduce armed forces along their common borders. Uzbekistan joined when the SCO was created in 2001, by which time terrorism, separatism and extremism had become its principal security concerns. The organisation did not begin as an alternative world order. It began because neighbouring governments had practical security problems that were easier to manage together.

Some of that machinery still works. The SCO created the Regional Anti-Terrorist Structure in Tashkent to connect member states’ security agencies, and cooperation has gradually spread into organised crime, narcotics and border security. In December 2025, representatives of SCO states participated in the Sahand-Antiterror exercise in Iran. This May, their anti-narcotics agencies met in Bishkek and agreed on draft programmes through 2030 on synthetic drugs, information exchange and coordinated operations. None of this resembles collective defence. It is bureaucratic cooperation: officials sharing information and procedures in areas where governments with otherwise divergent foreign policies still have overlapping interests.

The limits appear as soon as that cooperation becomes politically contentious. Counterterrorism is supposedly the SCO’s foundational issue, yet at the organisation’s defence ministers’ meeting in June 2025, the members could not even adopt a joint statement. India demanded language reflecting its concerns after the Pahalgam attack; consensus collapsed amid disagreement over terrorism and Pakistan. Reuters reported that the meeting ended without a declaration. The episode captured the SCO neatly. Security agencies can exercise together, but the organisation cannot make India and Pakistan agree on what terrorism means when the question touches their bilateral conflict.

That gap between technical cooperation and political consensus has widened as the SCO has expanded. India and Pakistan joined in 2017, Iran in 2023 and Belarus in 2024. Enlargement made the organisation geographically and symbolically larger while importing more disagreements into a system that operates by consensus. The SCO now represents around 25 per cent of global GDP and more than 3.4 billion people, but those figures tell us more about the size of its membership than its capacity to act collectively.

Why, then, keep joining?

India is perhaps the clearest answer. New Delhi participates in the SCO while also belonging to the Quad with the United States, Japan and Australia. It competes strategically with China, maintains an exceptionally close defence relationship with Russia and has a hostile relationship with Pakistan. SCO membership does not force India to reconcile any of these positions. It simply gives India another institutional space in which to manage all of those relationships.

Russia and China use the same room differently. Both have increasingly presented the SCO as evidence of a more multipolar international system and of institutions that do not depend on Western leadership. India is far less interested in making the SCO an anti-Western bloc. Yet the organisation can accommodate both positions because its members are not required to adopt a common grand strategy. The original SCO declaration explicitly described it as an organisation that was not an alliance directed against other states or regions. What once distinguished it from military blocs has become central to its durability.

Central Asia benefits particularly from that looseness. Kazakhstan, Kyrgyzstan, Tajikistan and Uzbekistan live with Russian security influence and rapidly expanding Chinese economic power, but none has an obvious interest in allowing either relationship to become exclusive. The SCO places both larger powers inside a multilateral institution in which the Central Asian republics are formally equal participants. It does not remove the enormous power asymmetry between them. It gives smaller states another arena to manage that asymmetry.

This is also why describing the SCO simply as a China-Russia vehicle misses something. Beijing and Moscow are indispensable to it, but an organisation designed only to transmit their preferences would be less useful to India or the Central Asian states. Kazakhstan can work with China through the SCO while expanding ties with Europe and the Gulf. India can sit beside China in Bishkek while strengthening Indo-Pacific partnerships elsewhere. Membership becomes an additional relationship rather than an exclusive geopolitical choice.

The Bishkek summit demonstrates the other practical benefit of such an institution: convening. The 31 August–1 September summit gives presidents and prime ministers a ready-made setting for bilateral meetings that might otherwise require their own diplomatic choreography. Rivals can share a summit without suggesting that their bilateral disputes have been resolved. For governments navigating several competing relationships at once, the availability of such spaces matters.

There is a price for all this flexibility. The SCO talks frequently about trade, connectivity and economic cooperation but has never created anything comparable to a common market or a meaningful economic union. It cannot guarantee collective defence. It cannot resolve the India-Pakistan conflict or the India-China border. Its expansion makes consensus still harder, while the increasingly ambitious language surrounding a new Eurasian or multipolar order regularly exceeds what its institutions can deliver.

But those failures become less puzzling once the SCO is judged against the job it can actually perform. It is strongest where states can separate a practical problem from their larger geopolitical disagreements: intelligence exchanges, anti-narcotics work, border contacts, meetings between officials. It becomes weakest precisely where cooperation requires members to agree on whose interpretation of a political conflict is correct.

Twenty-five years after the Shanghai Five began negotiating troops along a border, the SCO has therefore changed from a relatively focused security arrangement into something broader and looser. It cannot make India and Pakistan cooperate politically, reconcile India and China or turn Eurasia into a coherent economic bloc. It can keep their officials exchanging information, their leaders returning to the same summit and the Central Asian states from having to organise their regional diplomacy around one exclusive great-power patron.

That is considerably less than the architecture of a new world order. But it is also more useful than a talking shop. The SCO works because it asks its members to cooperate where they can and permits them to remain adversaries where they cannot.

By late summer 2026, the fight between Yemen’s Houthi movement and Saudi Arabia had stopped looking like a sideline to the bigger regional war. On 13 July, Saudi jets bombed the runway at Sanaa’s airport, apparently to keep an Iranian flight from landing there, and within a week the Houthis answered with a naval blockade of the kingdom and a run of strikes on its oil infrastructure. Between 13 July and 14 August they carried out sixteen attacks on Saudi targets, seven of them oil facilities and six of them tankers, hitting Abqaiq, the world’s largest crude-stabilization plant, and forcing Aramco to halt operations at Jizan. The campaign killed dozens of Saudi-backed Yemeni troops in Marib and Hadramout, and on 11 August it killed four crew members of a cargo ship transiting the Bab al-Mandab strait. By late August, the Houthis were still hitting Saudi-linked tankers off Yanbu, and their military spokesman had already declared the de-escalation phase of Yemen’s war over. Riyadh answered with a fourteen-country maritime coalition and its own strikes on Houthi positions, and a separate attack in early August left eleven Saudi civilians wounded. None of this came out of nowhere. It followed almost four years of quiet de-escalation that had held since the UN truce lapsed in October 2022, and it broke down precisely as Tehran found itself fighting a much bigger war with the United States and Israel. The obvious question is who is actually running this war. The honest answer takes you further than the battlefield, from Revolutionary Guard officers stationed in Yemen down to a financial holding company in Tehran that Reuters first mapped out thirteen years ago.

The immediate answer is the Islamic Revolutionary Guard Corps. Officers from the IRGC and from Hezbollah have been present in Yemen, helping direct Houthi attacks on shipping and supplying the targeting data that decides which vessels get hit. The Quds Force, through its deputy commander Abdul Reza Shahlai, oversees the flow of drones and cruise and ballistic missiles that turned the Houthis into a force capable of threatening the whole region, and reporting from inside Iran in 2025 claimed that Khamenei personally signs off on Houthi political and military decisions. By August 2026,Western and Gulf intelligence services were warning that the Houthis and Iraqi militias, working under IRGC coordination, might launch coordinated strikes on Saudi energy sites, ports, and airports at the same time. This is the Axis of Resistance doing what it was built to do: a set of armed groups managed by the Guard, answerable only to the Supreme Leader, used to project force that Iran cannot project on its own. Iran remains the Houthis’ main patron, supplying weapons, training, and intelligence, and the arrangement is close to a textbook case of what scholars call delegated deterrence: using non-state fighters as instruments of coercion so the costs of confrontation never land on Tehran directly. Yemen analysts increasingly rate the Houthis as a more useful partner for Iran than a battered Hezbollah, since they pair Iranian anti-ship missiles and long-range drones with a position straddling the Red Sea. The coordination is real but far from seamless. Since the war with Israel began, Tehran has tried to time missile launches from Iran, Lebanon, and Yemen together to overwhelm air defenses, even as Israeli intelligence penetration has repeatedly forced it to change how it does that.

But there is a second, financial answer to who controls this war, and it runs through the same man. The Supreme Leader who signs off on Houthi operations also controls a business empire that Reuters exposed in November 2013 in a series called “Assets of the Ayatollah.” The organization is Setad, short for the Execution of Imam Khomeini’s Order, founded in 1989 to manage property confiscated after the revolution, in theory for the benefit of the poor. Setad still describes itself that way, but the Reuters investigation found that under Khamenei it kept most of what it collected rather than passing it on. By 2013, Reuters put its holdings at roughly $95 billion: about $52 billion in real estate and $43 billion in corporate stakes, a sum around 40 percent larger than Iran’s entire oil exports that year and in the same range as the fortune of the Shah, who was overthrown in 1979. That empire was built by seizing thousands of properties belonging to religious minorities, including Baha’is, along with businesspeople and Iranians living abroad, often on the legal claim that the properties had simply been abandoned. Setad held a court-sanctioned monopoly on seizing property in the Supreme Leader’s name, and at one point auctioned off close to three hundred properties in a single month.

Setad’s relevance to the war in Yemen is indirect but decisive. Reuters found no evidence that Khamenei personally pockets money from the conglomerate, but through it he controls resources large enough to bypass parliament and the national budget entirely, which insulates him from factional politics and from rival branches of the Iranian state. The US Treasury sanctioned Setad in 2013, describing it as a sprawling network of front companies concealing assets on behalf of Iran’s leadership. Its revenues help fund the Leader’s compound and its staff, and just as important, they hand the IRGC a patron whose money does not depend on the state’s say-so. This is the material base underneath the proxy system: a leader with an independent war chest does not need permission from the government to arm the Houthis, and the IRGC, which reports to him rather than to any ministry, does the arming. The financial structure Reuters documented in 2013 is what makes the military structure visible in Yemen in 2026 possible in the first place.

Even so, the word proxy flattens what is actually happening, and the flattening matters if the goal is figuring out who is in control. The Houthis are closer to an informal partner of Iran than a puppet of it. They share Tehran’s ideology and its hostility to the regional order built by Saudi Arabia and the Gulf states, but they run their own leadership, hold their own territory, and pursue their own agenda. When the Houthis seized Sanaa in 2014, they inherited the conventional army of what had been North Yemen and now control territory holding roughly 80 percent of Yemen’s population, a combination of state institutions and geography that no other Iranian-aligned group can match. Iran’s approach across the Axis of Resistance gives each affiliate real latitude to chase local goals, which is why the relationship reads more like a strategic partnership than a chain of command running straight from Tehran. The Houthis themselves reject the label of proxy, and there is something to that: the blockade of Saudi Arabia serves grievances that are Sanaa’s own as much as it serves any Iranian strategy. A UN panel of experts has concluded that the Houthis could not build the weapons they now field without outside help, but that dependence has not translated into Iranian command over when and how they use them. Years of Hezbollah training have left the Houthis capable enough that they are now reportedly training allied fighters of their own, from Sudan to Iraq, a sign of how far the relationship has drifted from simple dependence. The collapse of the political track explains the timing: after more than four years of a de facto truce, negotiations with Riyadh stalled out, and Houthi leadership apparently concluded that military pressure on Saudi Arabia would get them further than talking would.

What makes this moment dangerous is that control and autonomy are running side by side, and each side is using the other. Iran reportedly asked its Houthi allies to act in its interest at the right moment, and an Iranian source has said that moment is now, with Tehran fighting the United States and Israel on a much larger front. With Hezbollah weakened and Hamas degraded, the Houthis have become Iran’s most capable remaining partner, and an Iranian ambassador along with a Quds Force representative have reportedly returned to Houthi territory to manage the relationship directly. But the Houthis are also riding Iran’s war to chase goals of their own, running what their spokesman calls a siege for a siege, meant to force Riyadh to lift its blockade of Yemen and negotiate on terms Sanaa can live with. Part of the strategy is clearly about establishing the Houthis as a regional actor in their own right, one capable of shaping events well beyond Yemen, not simply about supporting Iran. The group’s leadership has denied being anyone’s proxy and framed the campaign as a matter of Yemeni sovereignty, even while accepting Iranian weapons and coordination on the ground.

The war is directed but not owned. The IRGC supplies the weapons and the coordination, Setad supplies the money that keeps the whole system independent of the Iranian state, and the Houthis supply the will and the geography to fight. Each layer can push further than the layer above it would prefer. Saudi Arabia has responded with a maritime coalition and its own strikes, but neither has stopped the attacks, and the blockade has already forced Aramco to plan alternative export routes. The 2013 investigation into Setad was read at the time mostly as a story about one man’s wealth. It was really a story about how Iranian power is built. The Houthi war against Saudi Arabia is the same story, told this time in missiles rather than balance sheets, and it leaves an uncomfortable question hanging over the region: not whether Iran controls this war, but how much control anyone really has over a conflict where the commander, the banker, and the fighter each answer to a different master.

In the villages around Dawei, Myanmar’s latest development push has begun with soldiers. Since July, hundreds of troops have advanced through the area surrounding the long-stalled special economic zone on the Andaman coast, supported by air and naval units. Resistance fighters and local activists have reported burned villages, displacement and restrictions around the proposed industrial zone. Their accounts are difficult to verify independently amid active fighting, but the military’s priorities are clearer: shortly before the operation intensified, its deputy commander instructed forces in southern Myanmar to protect strategic projects. On 21 August, Reuters linked the offensive directly to efforts to revive the Russia-backed Dawei project, which includes plans for a deep-sea port and power station.

Dawei is an odd place to look for the next stage of Myanmar’s civil war, as the project predates the conflict by more than a decade. Myanmar and Thailand began pursuing a deep-sea port there in 2008. The plan was to combine a port, industrial estates and an overland connection to Thailand, giving Bangkok’s manufacturing economy a western outlet to the Indian Ocean and potentially reducing dependence on routes through the Strait of Malacca. The Myanmar Port Authority records the original 2008 agreement and the Thai developer’s subsequent failure to secure sufficient financing. Development plans were repeatedly revived, including through Thai and Japanese participation, but the port remained more convincing on maps than on the ground.

Finance was only one obstacle. Dawei also tested the political opening that began under President Thein Sein. Residents faced land loss, relocation and uncertainty over compensation, while local organisations complained that communities had little reliable information about a project expected to transform agriculture, fishing and village life. Research by EarthRights International documented disputes over consent, compensation, housing and livelihoods across affected villages. The politics surrounding a proposed 4,000-megawatt coal plant became particularly significant. Local campaigners distributed information, used the expanding media space and organised opposition; in January 2012, the government cancelled the plant, citing environmental concerns. A Cambridge study of advocacy around Myanmar’s special economic zones treats that campaign as evidence that organised communities had acquired some ability to alter large development projects.

The earlier Dawei struggle was hardly a story of communities controlling development. Land was still taken, compensation remained uneven, and residents rarely set the terms of the project. What they did possess was some capacity to contest those terms before they hardened into fact. Campaigns could delay construction, force environmental questions into public debate, and, in the case of the coal plant, alter what developers could build. In 2026, that space has narrowed further. The argument is no longer only about what kind of development Dawei should bring, but about who controls the territory on which any such argument can take place.

The February 2021 coup altered Dawei’s political context by reshaping Myanmar’s territorial character. Armed resistance spread well beyond the ethnic borderlands where insurgency had historically been concentrated. Ethnic armed organisations and newer resistance forces captured towns, roads and administrative space, while parallel forms of taxation and governance developed in areas outside effective military control. The humanitarian consequences remain enormous: the UN’s 2026 planning documents project more than four million internally displaced people and 16.2 million people requiring humanitarian assistance. The military’s election at the end of 2025 did not restore the lost territorial unity. According to International IDEA, voting was held in only 236 of Myanmar’s 330 townships, overwhelmingly in areas where the military could administer the process. The military-backed Union Solidarity and Development Party dominated the resulting parliament, which elected Min Aung Hlaing president in April. His opponents widely rejected the exercise, but its utility did not depend entirely on persuading them. It created a parliament, ministries operating under a formally civilian constitutional order, and a president whom foreign governments could receive as a head of state.

The election therefore offers a useful lens for understanding Dawei. The military did not need to restore uncontested sovereignty across all 330 townships before reconstructing some state institutions. Dawei applies a similar logic to economic geography. The government does not need to restore effective control across all of Myanmar before Dawei can function. It needs something more geographically limited: secure control over the port, its connecting roads and energy infrastructure, and the surrounding territory through which workers, goods and investment must move. A state that remains territorially fragmented can therefore still create economically viable corridors in the areas it can hold. These are corridors of governability rather than the recovery of uniform national control.

Russia has become a prospective partner in constructing one of them. Its post-coup importance to Myanmar began with arms, diplomatic protection and military cooperation, but the relationship has steadily widened. In February 2025, Myanmar and Russia signed an investment memorandum covering Dawei and surrounding areas. In June this year, Russian company Inter RAO and a Myanmar partner signed an MOU for a power station at the Dawei deep-sea port. A month later, Myanmar’s Commerce Ministry said procedures were under way to revive the SEZ and explicitly described Dawei as strategically important for the Thailand–Indian Ocean trade route. Those agreements should not be confused with a fully financed Russian takeover of Dawei. The project has disappointed too many previous promoters for another memorandum to be treated as a construction project. At this stage, Moscow is better understood as a prospective investor within an increasingly dense strategic relationship. When Vladimir Putin met Min Aung Hlaing on 18 August, discussions ranged from hydrocarbons and offshore exploration to a possible oil refinery and LNG supplies. Russia now offers Myanmar not only diplomatic backing but also the possibility of energy and infrastructure relationships that extend beyond the immediate requirements of war.

Naypyidaw has strong reasons to seek them. Myanmar’s economy remains far from recovery. The World Bank’s June economic monitor estimated that real GDP contracted by 2 per cent in FY2025/26 and that inflation reached 24.6 per cent in April. Electricity shortages, weak demand, rising costs and uneven implementation of government rules continue to constrain firms; many businesses remain focused on survival rather than expansion. Even the World Bank’s projected 2 per cent growth for the current fiscal year represents tentative stabilisation after a deep rupture rather than a restoration of the pre-coup economy.

A functioning port or industrial estate cannot reverse those conditions nationally. Its political value lies partly in its narrower scope. Dawei could create a visible enclave of investment, electricity, transport and foreign cooperation even while much of Myanmar remains economically damaged or militarily contested. The regime does not have to demonstrate that the national economy has recovered if it can point to places where long-term projects are restarting. The distinction between national recovery and investible enclaves matters because war itself changes the value of infrastructure. A village on a contested road is one kind of military problem. The same village beside a proposed port corridor becomes part of an asset whose security affects foreign investors, customs revenue and bilateral relationships. Territorial control becomes more valuable because capital is expected to follow it.

That does not make every operation around Dawei merely an infrastructure-clearing exercise. Tanintharyi has been an active theatre of armed resistance, and the military has independent reasons to regain routes and territory there. But the Dawei offensive shows counterinsurgency and economic planning beginning to converge on the same geography. Approximately 700 troops were involved in the recent push, according to Reuters, while the project’s revival was already under discussion with Russian partners and Thai officials. Investment is no longer imagined only after control is restored. The prospect of investment helps determine which spaces the state has particular reason to reassert control over.

Thailand makes Dawei more than a bilateral Myanmar–Russia story. Bangkok’s engagement with Naypyidaw follows from concrete economic dependencies that existed before the coup and survived it. Myanmar has been an important source of natural gas to Thailand for almost forty years; gas from the Yadana and Zawtika fields once supplied 15–20 per cent of Thailand’s total natural-gas demand, according to the Thai government. The two governments are now discussing extensions to gas contracts, new petroleum blocks and electricity-grid interconnection. For Thailand, relations with Myanmar therefore touch on electricity prices, industrial costs, and energy security at home.

Dawei would add another layer to that dependence. Its original economic logic was as much Thai as Myanmar: connect the Andaman coast to Kanchanaburi and onward to Bangkok’s industrial system. When Min Aung Hlaing arrived in Thailand for an official visit on 6 August, the relationship was no longer presented simply as crisis management at a difficult border. Thailand formally welcomed him as president, discussed trade, investment, energy and connectivity, and said it was ready to support Myanmar’s fuller participation in ASEAN.

It would be too simple to interpret that engagement as Thailand deciding that the coup no longer matters. Prime Minister Anutin Charnvirakul has instead called for “calibrated re-engagement” with Myanmar. Bangkok has also been involved in efforts to create space for talks as battlefield conditions shift. Earlier this month, Reuters reported tentative movement towards what might become talks about talks, with opposition actors signalling willingness to explore a political process even while the military had not recognised them as negotiating counterparts. Engagement may therefore serve mediation as well as commerce. But re-engagement also alters the sequence ASEAN originally envisaged after the coup. The bloc’s Five-Point Consensus presumed that reduced violence and political dialogue would pave the way for normal diplomatic participation. Myanmar’s government is now trying to reverse that order. Having held elections and installed Min Aung Hlaing as president, it has argued that ASEAN’s special envoy is no longer necessary, even though fighting continues and Aung San Suu Kyi remains detained.

The argument, then, is not simply that neighbours are abandoning democracy for business. Thailand cannot choose whether Myanmar exists on its border, supplies its energy or sends conflict and migration across it. Nor does speaking to Naypyidaw necessarily foreclose speaking to resistance actors. The more consequential possibility is that economic and diplomatic normality can accumulate before political normality does.

Dawei gives that accumulation a physical form. A port concession identifies the ministry authorised to sign it. A road identifies the authority expected to protect traffic. A power station creates an asset whose continued operation depends on specific security arrangements. A foreign investor begins planning on the assumption that those arrangements will persist. Infrastructure therefore does more than follow political order. When built during a civil war, it can embed one version of that order in land, contracts and cross-border relationships before a peace agreement determines how authority ought to be distributed. That is particularly important in Myanmar because the eventual political dispute is likely to be about federalism, not merely who occupies Naypyidaw. Ethnic armed organisations and newer resistance administrations have spent the post-coup years building their own institutions. International IDEA describes a country in which governance has become fragmented alongside military control. A future settlement would therefore have to negotiate not only elections but the distribution of fiscal power, natural resources, security authority and administration among multiple centres. Large infrastructure projects can pre-empt parts of that negotiation. A concession signed today can determine land use for decades. A gas or power arrangement can shape which level of government receives revenue. A corridor protected and taxed by one authority becomes harder to redistribute politically later. None of this means contracts signed by the present government will necessarily survive a settlement. It means the economic terrain over which federalism will eventually be negotiated is already shifting.

Dawei’s older history makes this especially clear. Communities around the project were never simply anti-development. EarthRights’ research on local views recorded demands for information, proper compensation, livelihood protection and meaningful participation. Residents could want roads and jobs while opposing the terms on which land was taken. The political question was not port or no port; it was who possessed bargaining power over the transformation that the port would bring. That bargaining space has shifted. During the reform period, even an unequal political system sometimes allowed communities to impose costs on developers and alter projects. In the present conflict, the question around Dawei is increasingly about who will control the land and guarantee access to it: which armed authority will do so? Once development becomes attached to a military corridor, a farmer resisting relocation, a resistance force contesting territory and a company seeking secure investment can all encounter the same landscape while asking entirely different questions of it.

Myanmar is therefore not moving through a conventional sequence from war to peace and then reconstruction. Elements of all three are now occurring together. The state is holding elections in territory it controls, seeking foreign investment in strategic corridors and rebuilding diplomatic relationships while fighting continues elsewhere. Resistance authorities are consolidating their own forms of governance. Regional states are experimenting with re-engagement even as formal peace mechanisms remain unsettled. The danger in describing this merely as “legitimising the junta” is that the phrase makes normalisation sound like a single diplomatic gift. What is happening is more gradual and material. Normality can be assembled through institutions and infrastructure: a presidential title, a bilateral energy agreement, an ASEAN meeting, a functioning road, a port concession. None of those acts ends the war. Together, however, they change the state that will eventually emerge from it.

Dawei matters because it shows how. The military-backed government is trying to make a contested territory investable before the country is politically settled. Russia offers a prospective partner; Thailand supplies an economic hinterland and an enduring energy relationship; the state supplies military control where it can. If the project advances, the port will not simply be evidence that Myanmar has begun rebuilding. It will become part of the institutional and territorial structure through which the government enters whatever political negotiation follows.

A port can be built before peace. The question is what happens when, by the time peace is finally negotiated, its roads, contracts, energy links and security perimeter are already shaping the country. Dawei once tested whether Myanmar’s political opening would allow communities to renegotiate development. Its revival now raises the reverse problem: whether development undertaken during civil war can begin to determine the political settlement that is supposed to come after it.

The remark wasn’t about Taiwan; that’s what makes it useful to Beijing

Anwar Ibrahim spent his Thursday (20th August) defending a position on Chinese reunification at a press conference after a TikTok Shop retail summit in Kuching. The Prime Minister launched the convention, presumably took questions on live commerce and small business onboarding, and then had to explain, again, why he told Al Jazeera in an interview with Sreenivasan Jain four days earlier that he’d use force to stop a Malaysian province from breaking away, and that Beijing was entitled to do exactly the same thing to Taiwan.

Sreenivasan Jain asks him, would he condemn China for reserving the right to use force on Taiwan. Anwar doesn’t think twice. He reaches for Malaysia itself as an example, a province deciding to break away, and says he’d do exactly what Beijing says it would do. Pressed on whether that means force, he doesn’t dodge it either. “Because otherwise they’ll be fragmented,” he says, which is about as close as a sitting head of government gets to saying that war might not be the last resort but the first choice.

Most countries that recognise Beijing over Taipei do it through the standard formula: acknowledge, note, respect, a whole vocabulary engineered to avoid saying anything with actual teeth in it. Anwar went straight to the premises underneath it. Not just that Beijing has the right, but that using it would be obviously correct, the same thing any government would do in the same situation.

China loved it, predictably. Lin Jian, at Wednesday’s briefing, said Beijing was pleased, grateful even, for what Anwar had said, then ran through the usual lines, inalienable territory, unstoppable historical trend. Global Times had an academic from Xiamen on by Wednesday evening, making the case that Anwar’s remarks prove the international consensus is already settled, and that any government with its own separatist headache should probably draw the same conclusion he did.

Taiwan’s foreign ministry took the opposite read, called it a threat to regional stability, and said it could cost Malaysia investment from Taiwanese businesses. The DPP called it false. None of it moved Anwar. By Thursday, he was standing by every word, pointing to the United States of America’s own history of refusing to let states secede, no country just lets a piece of itself walk out, and suggesting the whole thing was really about how Al Jazeera had framed the question rather than anything he’d actually said.

Which, technically, is true. Malaysia’s one China position isn’t new; it goes back to 1974, the joint communique under Tun Abdul Razak, Malaysia recognising the PRC as the sole legal government and Taiwan as part of its territory. Every government since has repeated some version of it. So Anwar’s defense holds up fine on the facts. What actually changed wasn’t the position. It was the volume.

And that’s the part worth sitting with, because the Taiwan line wasn’t really the point he was making. Go back to the interview itself. It comes up inside a longer answer about what Anwar calls Western hypocrisy, double standards on Gaza, and international law applied depending on who’s the one breaking it. Taiwan was the analogy he reached for to make a completely different argument, about consistency, about who gets to invoke territorial integrity and who doesn’t. He wasn’t working out a considered position on cross-strait sovereignty. He was making an analogy about the broken international order.

Which is, if we’re honest, roughly how China picks up most of its endorsements on this question. Nobody sits down and independently derives that Taiwan is a province of China. What happens instead looks a lot like this week: a leader mid-argument about something else reaches for an analogy, and the analogy happens to do Beijing’s work for it. Do that enough times, across enough interviews, across enough years, and you get what Lin Jian calls an unstoppable historical trend, which on closer look is really just an accumulation of exactly this kind of borrowed logic, tidied up afterward into something that sounds like consensus.

Anwar gets something out of the exchange too. China is Malaysia’s largest trading partner, and with this statement, an ever-grateful Chinese state might consider bolstering trade relations. It costs him nothing domestically, Malaysia doesn’t have a live secession movement worth the name, so the analogy was free to make and useful to have made. Whether that was calculated going in or just where his own rhetoric happened to land him, it’s hard to say from the outside. Probably doesn’t matter much to Beijing either way. They have got a new endorsement in the bank.

Zhu Rongji died this week in Beijing, 97 years old, and the Chinese papers are calling him all the respectful things now, statesman, revolutionary, loyal Party member till the end. None of that tells you much about what he actually did to the country. So let’s skip the tributes and go back to 1990 instead.

Zhu was on a tour of the US then, still just the mayor of Shanghai, and his American hosts kept calling him “China’s Gorbachev.” Meant as a compliment. He didn’t take it as one. “I’m not China’s Gorbachev,” he told them. “I’m China’s Zhu Rongji.”

Fair question, why did that comparison bother him so much. Gorbachev is shorthand for one specific kind of failure, an economy you open up faster than the political system underneath it can take, and the whole thing comes apart. For the next thirty years China’s entire bet was that this particular failure could be avoided, that you could let the market in and keep the Party’s grip on everything else. Zhu is a big part of why people believed that bet could work. And a big part of why, for a long time, it did.

He got there by a strange enough route. Orphaned young. Trained as an engineer. Denounced twice over, first as a “rightist” in the late fifties for saying Mao’s growth targets were irrational, which cost him his party card and five years doing menial labour in the countryside. Purged again in the Cultural Revolution, another five years, same story. Nobody rehabilitated him till 1979, once Deng was firmly in charge and pragmatism was briefly back in fashion. After that the rise was quick. Shanghai mayor in 1988, party secretary a year later. And it was as party secretary, in June 1989, that he did the thing that tells you almost everything about how he’d govern for the next fifteen years. Beijing was calling the massacre in Tiananmen “turmoil.” Zhu published something in the Shanghai papers that used neither that word nor any softer one, just that what had happened was a fact and facts don’t stay buried. Not defiance exactly. Not the official line either. Just the first sign of a talent he’d spend the rest of his career on, saying the true thing in a way the Party could still swallow.

By 1991 he was vice premier, and when Li Peng’s health gave out in 1993 he took the whole economy over. Two moves right away. As de facto central bank governor he cut off the credit that was pushing inflation towards 25 percent, and had it down to zero within two years, over howls from every state enterprise boss who’d gotten used to easy money. And separately, on the tax side, he centralised revenue collection so hard that provinces went from keeping most of what they raised to sending most of it to Beijing. That 1994 reform is probably the single biggest thing Zhu Rongji ever did. Worth remembering, because we’re coming back to it. It’s also the reform whose bill has come due.

But first, 1998, when the premiership he’d been circling since ‘93 finally landed, and he opened with a promise that was reckless even for him: fix the state banks in two years, fix the state enterprises in three, “minefield or bottomless abyss,” didn’t matter, he’d see it through. The banks took closer to ten. The enterprises never really got fixed, not the way he meant. Getting there meant laying off something like thirty million state workers, most of them in the old industrial northeast, and the welfare net he built to catch them was thin and mostly improvised. It didn’t catch everyone. Parts of that region never came back.

What he did land, on the outside-facing side, was the WTO. Six and a half years of negotiation, much of it face to face with the American trade rep Charlene Barshefsky, ending in December 2001 with China inside the global trading system just in time for a decade of double-digit growth. Barshefsky’s line on him, years later, was that he genuinely wanted China to look more like the West, more open, more market-run, bound by rules it hadn’t written alone. Someone at Goldman Sachs who dealt with him back then compared him to Ludwig Erhard, the West German technocrat who built that country’s postwar economy out of similar rubble, which is a big claim but not a crazy one.

He was, whichever way you look at it, a strange fit for running a one-party state. Loved sparring with reporters instead of reading off cards, which almost nobody at his level does. Fired an official on the spot once for showing up with a cigarette lighter he couldn’t have afforded on his salary. Had a favourite line, depending who’s telling the story, about a hundred bullets or a hundred coffins kept in reserve, ninety-nine for corrupt officials, one for himself. It didn’t win him friends inside the Party. He never built the patronage networks that might have gotten his pension reforms or his bad-debt cleanup through, and both quietly died on the table. He made enemies faster than allies and, from what I can tell, didn’t much care.

Here’s the part most of the tributes are skipping past. That 1994 tax reform, the one that made Beijing rich and left the provinces broke, it didn’t just fund the central government. It also left every province and city with exactly one lever left to raise money on their own: land. Sell it, lease it, borrow against selling more of it later. Three decades of Chinese growth ran, in part, on that lever. And so, it turns out, did the property slump China is still stuck in, and the something like 100 trillion yuan in local government debt that Xi’s government is now trying, slowly and half-heartedly, to unwind. Standard Chartered’s economists actually call Xi’s 2024 fiscal overhaul the third big tax reform of the reform era (Fortune, 2024), the first being Zhu’s, in 1994. Which is a polite way of saying Beijing just spent two years trying to fix a problem Zhu built into the machine thirty years ago.

None of this is really an argument against him, to be clear. You can’t blame someone laying pipes for a trillion-dollar economy in 1994 for not planning around the strains of one twenty times the size. But that’s the honest version of “the China he shaped.” Not a clean inheritance. A machine that’s still recognisably his, still running on the incentives he wired into it, debt and all.

Barshefsky said one more thing, the less flattering one, and it’s probably closer to the real epitaph (NPR, 2026). Zhu, she said, belonged to a China from a different time, one that wanted to become more like the West. Xi’s China wants something else, more state, more self-reliance, less of the outside world let in. Zhu spent his last years quiet, in a house in the hills west of Beijing, listening to Peking opera, and apparently still advising some of the younger technocrats running the place. Not all of them were listening, or so people say. Maybe that’s the real question here. Not whether China still needs someone like him. Whether it could still make room for one.

A hall in Shanghai, the evening of 16 July 2026. Chandeliers, a long table, twenty-nine foreign ministers and deputy prime ministers waiting their turn with a pen. The UN Secretary-General sits in the front row and not signing, just watching. China’s foreign minister, Wang Yi, signs first. By the time most of the world wakes up to read about it, the World Artificial Intelligence Cooperation Organization already exists, headquartered in Shanghai, its founding text committing every signatory to “the purposes of the UN Charter” and a “people-centered approach.” Guterres himself attended, lending the ceremony a legitimacy no press release could buy on its own. The deputy prime ministers of Kazakhstan, Laos, and Pakistan sign for their governments. Nobody rushes.

The next morning, Xi Jinping takes the stage at the World AI Conference and delivers the line that will end up in every headline: AI development, he says, should not be “a solo performance by a single country” but rather “a symphony of international cooperation.” It’s a good line. It’s also, notably, aimed at an audience that isn’t in the room, because the room, for all its ceremony, is missing some conspicuous names.

No G7 state signed. No EU member. Not the United States, not the UK, not Japan, not Canada, not South Korea. And not India, which currently holds the BRICS presidency and is, pointedly, the only one of the five founding BRICS members absent from the founding roster. A body built to speak for the developing world just launched without the developing world’s largest democracy at the table.

Whether WAICO is admirable multilateralism or soft-power theatre dressed in UN language is a debate for people who enjoy debates. The sharper question, and the one worth actually answering, is this: what did twenty-nine countries just sign up to receive?

WAICO is a standing intergovernmental organization with a Council and a Secretariat, its mandate spanning capacity-building, technical standards, interoperability, open-source ecosystems, and coordination with existing UN processes, confirmed both by the Observer Research Foundation and by India’s World’s own explainer. It sits deliberately outside the UN system, a design choice, as Caixin’s coverage frames it, not an accident of drafting.

Xi pledged 5,000 AI training and seminar places for people from developing countries over the next five years, continued open-source model releases, and new cooperation centres with ASEAN, the African Union, and BRICS. That’s a number, a timeline, and named partners, which are more concrete than most diplomatic communiqués manage.

The membership rules matter as much as the promises inside them. Anyone can join. No values test, no regime-type screen at the door, the whole architecture is built around development and the global capability divide rather than shared politics. Compare that to the alternatives on offer: the G7’s Hiroshima Process and the American-led Pax Silica initiative are, at bottom, clubs of like-minded states. WAICO deliberately isn’t. Arindrajit Basu of the Carnegie Endowment puts the strategic logic bluntly: with Washington pulling back from AI norm-setting, he says, Beijing is “keen to demonstrate its global leadership” and it’s betting the pitch lands hardest with the roughly 130 countries that are neither chip designers nor chip fabricators, but will make up AI’s largest user base within a decade.

The lineage matters too, because it kills the lazy read that this was thrown together as a fast reaction to a geopolitical opening. It wasn’t. The Global AI Governance Initiative launched in 2023. An AI capacity-building action plan followed in 2024. Li Qiang proposed the organization by name at the 2025 WAIC. Xi reiterated it later that year. The signature came in July 2026. Three years, four deliberate steps. Whatever else this is, it isn’t opportunism.

Same government, same month, different ministries, and this is where the story actually earns its interest. On 7 July, Reuters reported, independently confirmed by TIME and Yahoo, that Chinese regulators, led by the Ministry of Commerce, had met with Alibaba, ByteDance, and Zhipu to discuss restricting overseas access to the country’s most advanced AI models, including systems not yet released. A tiered regime was floated: light filing for less capable open models, formal security review for stronger ones, and a possible ban on public release for the most capable of all. Officials also raised treating the theft or leak of proprietary AI as a national-security offence, which is the kind of legal reclassification that turns a corporate dispute into a criminal one.

Morrison Foerster’s reading of the State Council’s new Regulations on Outbound Investment, which were published on 1 June and came into force a month later, confirms that Article 13 folds technology transfers conducted through deploying technical personnel overseas, or arranging cross-border training, squarely into export-restriction territory. Not just formal IP assignments or licensing contracts. Training itself.

This produces an irony almost too neat to be accidental: WAICO’s flagship deliverable is training 5,000 people from developing countries, and China’s own outbound-investment rules, issued the same month, tighten the screws on technology transfer conducted through exactly that kind of training arrangement. Xi’s own keynote warned other governments against “overstretching the national security concept” in AI, delivered in the same month his government was busy overstretching its own.

There’s a fair version of Beijing’s position, and it deserves an honest hearing rather than a dismissal. Exporting diffusion while guarding the frontier isn’t hypocrisy; it’s the same instrument the United States has been running, just pointed in the opposite direction along the same stack. Morgan Lewis’s analysis of the BIS rule that took effect 15 January 2026 confirms the shift from a blanket presumption of denial to case-by-case licensing for chips under specific performance thresholds, roughly H200/MI325X-class hardware, conditioned on guaranteed US-market supply first, a hard 50-percent cap on the volume share that can go to China or Macau, mandatory pre-shipment testing by an independent US lab, and know-your-customer obligations on the receiving end.

Both powers have landed on the same design: access to capability at one tier, tight licensing at the next, the gap between the two held as leverage. China has simply bolted an intergovernmental development body onto the front of that structure, something Washington, notably, has not done. White House AI policy advisor Sriram Krishnan has said the quiet part out loud on the American side: America wants “the world” to build on “the American AI stack,” chips and all. Beijing’s pitch is less blunt but not fundamentally different in structure, just dressed in different language and delivered with a training programme attached. Observer Research Foundation’s Basu Chandola calls the strategy “selective shaping” and adds that Chinese openness is best understood as openness to the diffusion of Chinese capabilities rather than open access in any broader sense.

The consequence for WAICO’s members is where all of this actually lands. Capacity-building delivers the use of AI: applications, trained people, and working infrastructure. It does not obviously deliver the ability to build independently, not if the tiered export-control proposal now sitting in Beijing’s consultation process eventually gets adopted. In that scenario, weights, training pipelines, and inference infrastructure would cross borders by licence, not by download, the way most of them do today. A member state could become a genuinely competent operator of Chinese AI without ever becoming a producer of it, running the models fluently, training its people on them, building a domestic services layer on top, while never quite holding the keys to reproduce the stack or negotiate the terms of its next generation. But that whole outcome hinges on an export-control package that, as things stand, remains a proposal under consultation rather than settled law.

India’s stated reason for staying out is security concerns; the broader state of the relationship with Beijing is legible and defensible on its own terms. But staying out is only a strategy if the alternative is properly funded, and that story turns out to be more complicated than a single number suggests.

India budgeted ₹2,000 crore (roughly ₹20 billion) for the IndiaAI Mission in FY 2025–26. According to MediaNama’s reporting on a Rajya Sabha reply from MeitY, dated 9 February 2026, two things happened before a rupee moved. First, the government’s own revised estimate for the year quietly slid from ₹2,000 crore down to ₹800 crore. Second, actual disbursement against that already-lowered figure came in at ₹379.15 crore. Measured against the original ₹2,000 crore line, that’s roughly a fifth, an alarming underspend on its face. As Observer Research Foundation’s Sameer Patil put it on the broader dependency question, access “can be shut down overnight”, a warning that cuts as sharply against India’s own underfunded sovereign push as it does against reliance on any foreign model.

The structural problem sits above the budget line entirely. Beijing intends to expand WAICO’s reach through BRICS and the SCO, two forums where India holds full membership, and where India currently chairs the former, with the eighteenth BRICS summit due in New Delhi later this year. New Delhi is set to negotiate around WAICO’s agenda from inside the room next door, while holding no seat at WAICO’s own table. Adding to the regional picture: five ASEAN states signed on as founding members, Indonesia, Malaysia, Cambodia, Laos, and Myanmar, alongside Kazakhstan, Kyrgyzstan, Tajikistan, Uzbekistan, and Oman. On India’s own periphery, the Chinese framework already has formal institutional standing. India, for now, has a parallel summit track and little else.

India hosted the 2026 AI Impact Summit, part of the safety-and-standards lineage running from Bletchley through Seoul and Paris, drawing endorsements from more than ninety countries and organizations, the United States, China, the UK, Russia, and Brazil all among them. India has also signed onto Pax Silica. That track carries genuine legitimacy and real Western buy-in. Whether it has much to offer a finance ministry in Lusaka weighing its options is a considerably harder question, and one nobody in New Delhi has fully answered yet.

An important takeaway for the member states, not for Delhi and not for Beijing, is that they seem to have signed a development agreement in the exact month the host government was quietly negotiating an export-licensing regime with its own leading AI labs. Both processes are public and traceable. Any state that joined WAICO on the strength of its capacity-building promises should read the 16 July founding agreement alongside the MOFCOM consultation that’s been unfolding in parallel through late July, and price the offer accordingly. The door is still swinging open: Bangladesh announced on 1 August that it will join as an observer, which is the latest in what looks like a steady trickle rather than a single closing ceremony.

The twentieth-century development bargain ran on concessional finance with policy conditions attached. The bargain now taking shape is assuming Beijing’s export-control consultation lands where the tiered proposal currently points and would run on concessional capability with conditions attached to the export licence instead: terms set in a ministry’s guidance document rather than a loan covenant, just as easily revised without renegotiation the next time the frontier moves. That’s a plausible trajectory based on the evidence gathered so far. It isn’t yet a fact on the ground, and the space between those two things is exactly where the next year of reporting on WAICO needs to live.

Thailand’s latest attempt to revive its economy arrives in two halves. One is meant to soften higher fuel and living costs through consumer support. The other is supposed to finance a cleaner energy system that would make the country less vulnerable during the next oil shock. On 9 July, the Constitutional Court cleared Prime Minister Anutin Charnvirakul’s government to borrow 400 billion baht for both. The accounting is neat; the economic problem is not. A payment can ease next month’s bill. An energy transition must survive several governments.

Before the ruling, Rungthiwa Pimphanit was waiting for an earlier promise of immediate movement. The government employee in northeastern Nong Bua Lam Phu had expected Pheu Thai’s 10,000-baht digital-wallet payment to cover her son’s school supplies. Tens of millions of people had been promised money that had to be spent near home and within six months, pushing cash through local shops before it could settle into savings. By June 2025, Rungthiwa was still waiting and said she would not vote for the party again. When the wallet stalled, its unspent money revealed what the promise had obscured. After distributing only part of the programme, the government redirected 157 billion baht towards water systems, infrastructure, productivity measures and local businesses. Money designed to circulate had ended by financing the systems that allow circulation to become income. The reversal did not show that households needed no relief. It showed that spending and productive capacity were never interchangeable.

The same distinction runs through the political tradition from which the wallet emerged. Yingluck Shinawatra’s first-car tax rebate encouraged households in 2011 and 2012 to bring forward a major purchase. Vehicle sales rose, but research using national credit-bureau records later found higher loan delinquency among participants and less access to other borrowing. The car arrived early; the instalments remained. A decade earlier, Thaksin Shinawatra’s 30-baht healthcare programme had also moved with political speed. Introduced in 2001, it helped Thailand achieve universal health coverage in 2002 by extending protection to people excluded from existing schemes. The announcement passed; the institution remained. One programme shifted demand from the future into the present. The other changed what citizens would encounter in the future.

That difference has become more consequential as Thai households and firms lose room to absorb another accelerated purchase. Household debt stood at 86.7 per cent of GDP at the end of 2025, while more than half of workers remained in informal employment. Relief entering such an economy meets vehicle instalments, mortgages, farm loans and incomes too uncertain to support affordable credit. A household may sensibly use a payment to keep up with old obligations. A small shop may enjoy a temporary rise in sales without gaining the finance, equipment or trained workers needed to produce more the following year. None of this makes Thailand an incapable economy. It remains a sophisticated centre for manufacturing, exports and tourism. The difficulty is that these strengths no longer translate easily into broadly rising incomes. The IMF now expects growth of only 1.9 per cent in 2026. When the slowdown comes from weak productivity, uneven firm capability and indebted households, another burst of consumption treats the point at which the problem becomes visible rather than the machinery producing it.

Economics alone, however, cannot explain why the same remedy keeps returning. Since Thaksin’s removal in 2006, governments and successful parties have repeatedly been interrupted by coups, court decisions, dissolutions and unstable coalitions. Reforms to education, competition or industrial capability may mature under somebody else’s administration. A transfer reaches an identifiable beneficiary before the next rupture. Rungthiwa revealed the other side of that bargain: once relief becomes an electoral contract, voters remember precisely who failed to deliver it.

The July loan returns Thailand to this choice. Immediate support is defensible because families facing higher food, fuel or fertiliser costs cannot live on a promise of future productivity. Yet the subsidy will be visible first. The energy transition will require grids, investment, regulation and years of implementation after prices stabilise. Borrowing can bridge the interval between the two, but it cannot repeatedly substitute for what is supposed to happen during that interval.

Thailand’s warning to other slow-growing economies is not against acting quickly. Its universal health system demonstrates how rapid action can create lasting capacity. The test is what remains when the immediate benefit has been spent. The car rebate left repayments. The digital wallet left disappointed voters and a budget redirected towards the infrastructure it had initially tried to bypass. The present loan may leave a less vulnerable energy system, but only if its quieter half survives the political success of the subsidy. A quick intervention buys time for change. A quick fix asks the next payment to buy the same time again.

Before hostilities escalated in the vital Middle Eastern chokepoint, 110 vessels a day transited the strait carrying crude, refined products and LNG. By late February, that figure had collapsed to near zero. A mid-June deal briefly restored traffic to about 35 crossings a day, still a third of normal, before a second, harder escalation cycle beginning around 7 July pushed it back down: by 23 July commercial transit had fallen to roughly 11 per cent of pre-crisis volume, ten vessels against a typical eighty-eight. The strait did not reopen so much as flicker. In an ordinary year the passage carries close to a fifth of the world’s oil and gas shipments, and most of it moves east.

Two Asian economies received this shock at the same time and did entirely different things about it.

That exposure was never symmetric to begin with. Western economies absorb the same kind of shock through Atlantic Basin supply and domestic production that Asia does not have. India’s own exposure had actually been rising before the war: by March 2026 around 52 per cent of its crude imports were passing through Hormuz, up from 41 per cent the year before, as refiners had trimmed Russian purchases in favour of Gulf term contracts. China’s dependence sits higher still: close to 89 per cent of Iran’s own crude exports go to China alone. The LNG exposure matters more quietly, but it is the one that carries through to the Japan section later. In 2024, about 20 per cent of global LNG trade moved through Hormuz, and 83 percent of that was Asia-bound, with China, India and South Korea together taking 52 percent of it; Europe’s equivalent exposure is a fraction of that. Bypass routes exist: Saudi Red Sea terminals, Fujairah, the Iraq-Türkiye pipeline to Ceyhan, Iran’s own Jask terminal, but the April assessment from the IEA and Kpler showed those routes absorbing only a fraction of what the strait normally carries, with weeks added to transit time in the bargain.

India’s answer was arbitrage, and the shape of it is collapse, then capture. Indian fuel exports fell to about 866,000 barrels a day in May, the lowest in nearly four years. Hindustan Petroleum’s managing director said the company received almost nothing against its Middle East term contracts in the first quarter, because the cargoes sat stranded on the far side of the strait. Indian Oil abandoned a plan to send the supertanker Lila Jamnagar, a two-million-barrel vessel, through Hormuz, judged the crossing not worth the risk, and suspended Iraqi liftings along with Mangalore Refinery. Then came the recovery. July is tracking near 1.55 million barrels a day of light and middle distillates, close to double May and the second-highest month in Kpler’s series going back to 2017, on the back of the ceasefire’s collapse and the refining-margin rally that followed. Refiners tested new grades from Angola and Venezuela while Russian volumes stayed near record levels through July.

This could be viewed as a triumph, but it isn’t one. The capacity is being monetised: Jamnagar, Vadinar, the coastal export complexes were built over two decades for entirely commercial reasons, long before anyone in Delhi was thinking about a Hormuz contingency. The state did not plan this. It was positioned for it, which is a different claim. The comparators keep the claim honest: Oman, shipping from ports outside the strait entirely, hit an all-time high of 783,200 barrels a day in July; Taiwan reached 306,000 barrels a day, its strongest month since December; and Asia’s total distillate imports, even with all of this, sat at roughly 5.80 million barrels a day in July, up from June’s five-year low but still 18 percent below the 7.05 million barrel average of the three months before the war. India is the largest single filler of a gap that is still open.

India’s state refiners went into July holding roughly two months of crude cover, and were in no hurry to resume Gulf term liftings even once the strait showed signs of reopening. None of this is autarky or stockpiling. It is a state controlling the pace of its own re-entry into a market it does not control the entry point to. In an import-dependent economy, its capacity shows up less as reserves or pipelines than as the ability to choose when to normalise.

Japan’s answer was the opposite instrument entirely. Defence minister Shinjiro Koizumi said in a webcast in July that the country could no longer treat the subject as unspeakable: “While this is a difficult issue for Japan to discuss, one topic we cannot avoid is nuclear weapons,” he said. Koizumi pointed to precedent outside Japan, Macron’s March announcement of an expanded French arsenal, and Finland’s parliamentary repeal of its own nuclear prohibition in June. Sergei Lavrov, speaking in Manila after the ASEAN foreign ministers’ meetings on 22 July, called Takaichi’s statement “a serious signal,” and said it echoed drills already underway between the United States and South Korea.

The Northeast Asian latency debate is not driven only by doubt about American extended deterrence but also driven by energy. Japan and South Korea import nearly all of their oil and LNG. Hormuz hit both directly, and the pressure to secure stable supply feeds straight into the argument for domestic nuclear capability, civil first and latent after. The taboo is eroding precisely where energy insecurity and deterrence insecurity meet. Japan’s timeline advantage over South Korea is worth a line and no more: nuclear expert Siegfried Hecker has assessed that Japan could assemble a uranium weapon in six months to a year, a plutonium weapon in up to two, against a longer timeline for South Korea’s fissile material production alone. That is a fact about capability, not a forecast about intent.

Comparison here holds up best as a mechanism, not a moral. Two variables do the explaining. First, the substitutability of inputs. India can swap one crude grade for another because it has refining flexibility and roughly forty suppliers to draw on. Japan cannot swap a nuclear umbrella for a different umbrella. Second, the autonomy of the security guarantee itself. India’s deterrent is its own, and its Hormuz exposure is painful and commercial. Japan’s energy exposure is close to existential, and the deterrent meant to cover its worst cases belongs to someone else. India hedged downstream, in the part of the chain it actually controls. Japan is hedging upstream, at the level of the guarantee itself, because that is where its exposure sits. Neither state chose its instrument freely. Each reached for the only one within reach. A shock does not reveal a state’s intentions. It reveals its inventory.

On 28 July the US Senate voted 86 to 12 to advance the Lindsey O. Graham Sanctioning Russia and Iran Act of 2026, which would let the president impose tariffs of up to 100 per cent on the world’s five largest buyers of Russian crude, currently China, India, Slovakia, Hungary and Azerbaijan, with a separate and more forgiving threshold shielding most European gas buyers. The bill requires the top-five list to be reassessed every 180 days, and still needs a House vote and a presidential signature before it becomes law.

So here’s the asymmetry. India’s arbitrage runs, in part, on Russian crude that a foreign legislature revisits twice a year. Japan’s hedge, if it proceeds, will not need anyone’s recertification. Both are bets on autonomy. Only one of them answers to somebody else’s calendar.

There is a body in Beijing that most people outside China have never counted, and it is worth counting, because the arithmetic is the whole story. The Politburo of the Chinese Communist Party has twenty-four seats. Not twenty-four hundred, not two hundred, twenty-four. These are the people who run the second-largest economy on earth and the largest army, and they meet in rooms the rest of us never see, and their names are known to specialists and almost no one else. It is a small number. When one of them falls, the number is smaller, and you can feel the change the way you feel a missing tooth with your tongue.

On the fourteenth of July, the party expelled Ma Xingrui. He had been the party secretary of Xinjiang, the vast western region, and before that the governor of Guangdong, and before that a space engineer who helped run the country’s programme to put men in orbit. He was, by the standards of the system, a serious man with a serious record. The anti-corruption watchdog said he had traded favours in the appointment of officials, taken bribes, arranged jobs for people he should not have, and, according to one report, exchanged political favours for sex. The language the watchdog used was, in the words of a scholar at Zurich who reads these notices for a living, extremely serious, with some highly unusual charges. When the party wants to bury a man it has a vocabulary for it, and the vocabulary used on Ma was from the harsher end of the shelf.

He was the third member of the Politburo to be brought down in this cycle. That is the fact worth sitting with. The first two were soldiers. Zhang Youxia, the most senior general in the country, was placed under investigation in January. He Weidong, a vice chairman of the military commission, was expelled last year, which took the body from twenty-four down to twenty-three. Both were military men, and there is a way of reading military purges that lets you sleep at night, if you are the kind of person who worries about the stability of China. Armies are dangerous to those who rule them. A leader who purges his generals is doing something old and comprehensible, which is making sure the men with the guns do not get ideas. You can file it under self-preservation and move on.

Ma is harder to file. Ma was a civilian. He ran provinces, not divisions. He was an administrator, a technocrat, a man of the bureaucracy rather than the barracks. And when the purge reaches past the soldiers and into the civilian administration at the very top, the reading changes. It is no longer only about the guns. It is about the whole apparatus, the entire class of people who run the country, and the message it sends to every official below them, which is that no rank is high enough to be safe and no record is clean enough to be a shield.

I want to be careful here, because there are two stories you can tell about a thing like this and they point in opposite directions, and honesty requires holding both before choosing.

The first story is that this is exactly what it says it is. China has a real corruption problem, the kind that rots public trust and misallocates enormous sums, and Xi Jinping is doing the hard and unglamorous work of cleaning it out, floor by floor, without regard for rank. In this telling the purge is a sign of health. A system that can discipline its own leaders is stronger than one that cannot. Most states, faced with corruption at the top, look away, because the people who would have to prosecute are the people who would have to be prosecuted. China, in this account, is doing the thing that is supposed to be impossible, which is holding its own princes to account.

The second story is quieter and colder. It is that anti-corruption, in a system with no independent courts and no free press and no opposition, is not a legal process but a political instrument. There is no way, from outside, to know whether Ma Xingrui was guilty of what he was accused of, or guilty of something else, or guilty of nothing beyond ending up on the wrong side of a calculation being made above him. The charges are announced by the same power that decides who is charged. When there is no daylight between the accuser and the judge, the accusation tells you about the accuser’s intentions and very little about the accused’s conduct. In this telling the purge is not about corruption at all. It is about control, and the word corruption is the coat it wears in public.

Both stories are probably true at once, in proportions no outsider can measure, and that is the uncomfortable heart of it. Ma may well have done what they say. Powerful men in unaccountable systems very often do. But the fact that he was caught, and when, and how the charges were framed, and that the decision to end a man of his rank was taken at the very top with Xi’s own backing, all of that belongs to the story of control, not the story of law. A guilty man can still be a convenient one. The two are not in tension. They fit together like a hand in a glove.

What the third chair tells us, the civilian chair, is that the machinery has become permanent. This is the part that matters more than any single name. In the early years you could describe the anti-corruption campaign as a campaign, a thing with a beginning and a middle and an eventual end, a housecleaning after which normal life resumes. Fourteen years on, it has not ended, and it shows no sign of ending, and it has its own standing bureaucracy, the Central Commission for Discipline Inspection, with its own investigators and its own procedures and its own grim momentum. It is no longer an event. It is a feature. It is how the system now works, a permanent engine of discipline humming under the floor of Chinese political life, and every official at every level lives with the knowledge that the engine could turn toward them next.

There is a cost to running a state this way, and it is not the cost you would expect. The obvious worry is instability, the fear that a leader who purges this widely will eventually purge someone who fights back. But that is the less interesting danger. The deeper cost is quieter. When the price of prominence is the permanent possibility of ruin, the capable and the cautious begin to behave in a particular way. They stop taking initiative. They stop making decisions that could later be read against them. They wait for instructions and they cover their traces and they push every choice upward, because a choice made is a choice that can be punished, and a choice deferred is safe. A government of frightened administrators is not an efficient one. It is a government where nothing moves until the centre commands it, and the centre, however clever, cannot command everything.

So the purge that looks like strength contains a weakness, and the weakness is not that the system will crack. It probably will not crack. The weakness is that it will stiffen, that the very machinery built to keep the officials honest will also keep them still, and that a country which needs its administrators to be bold and inventive at exactly this moment, with the economy slowing and the old growth model spent, will find them instead crouched and waiting, watching the engine under the floor, listening for the sound of it turning their way. Ma Xingrui was the third to fall. The men who are still seated learned something on the fourteenth of July, and what they learned was to be careful, and careful men do not build the future. They survive the present. It is not the same thing.

Xi Jinping stood in Shanghai’s World Expo hall on July 17th and made his pitch in one line: “人工智能发展不应该是某个国家的独奏，而应当是全球合作的交响” - AI’s development, he said, should not be one country’s solo, but a global symphony. Kazakhstan’s Tokayev was in the room. So were Cambodia’s Hun Manet, Thailand’s Anutin, and UN Secretary-General Guterres, with delegations from more than a hundred countries there to watch China sign the World AI Cooperation Organisation into existence. Four days earlier, China’s own statisticians had confirmed the number the keynote never mentioned: the economy grew 4.3 per cent year-over-year in the second quarter, missing the consensus forecast of 4.5 percent and marking the weakest quarterly expansion since the country was still grappling with zero-Covid in late 2022. Beijing wrote this score alone, in a room without the other players, and is now asking the rest of the world to play from it before anyone checks it against the state of the band.

Here is the arithmetic. Fixed-asset investment fell 5.7 per cent in the first half, worse than forecast, dragged down by an 18 percent collapse in property investment. Retail sales grew a bare 1 per cent in June, the third straight soft month, as households pull back rather than spend. Exports rose 27 per cent year-over-year in June, and integrated-circuit shipments and AI-related equipment carried almost all of it. Everything domestic is contracting; the one line still growing is the one built on chips and AI export hardware.

Caixin’s economists caught something sharper than the headline miss: China’s nominal GDP growth beat its real GDP growth for the first time in thirteen quarters. Demand didn’t recover; AI-supply-chain prices and oil prices pushed the GDP deflator positive instead. The only part of the Chinese economy still able to set a price upward this quarter is the part built on chips and AI hardware. Cornell’s Eswar Prasad said it without the diplomacy attached: “China’s economy is becoming increasingly unbalanced,” with investment piling into frontier technology while consumption stays flat. The imbalance is the reason for the Shanghai summit.

This is state control, not market momentum. Beijing directs capital into AI and semiconductors by command, the same way it ordered the food-delivery discount war to stop last year, and a World AI Cooperation Organisation built by a state that runs its own economy this way carries that control with it into the framework it offers everyone else. Five thousand AI training slots for developing-country officials over five years. Joint application centres built with ASEAN, the Arab League, the African Union, CELAC, the SCO and BRICS. A weather-forecasting system already running in thirty countries. Each of these is genuine capacity-building for countries that need it, and each also converts a domestic dependency into an export relationship on Beijing’s terms before anyone else gets to define the exchange. A government this size doesn’t project confidence abroad to hide anxiety at home; it manages the anxiety at home by making the rest of the world need what it sells.

Delhi ran its own summit five months earlier, one China skipped: roughly seventy countries moved toward a Delhi Declaration with Modi, Sam Altman, Sundar Pichai and Macron in the room, the same week a White House official told the room outright, “We totally reject global governance of AI”.

The gap Delhi can’t out-declare is technical, not rhetorical. Chinese open-weight models like DeepSeek and Qwen have logged more than ten billion cumulative downloads and, by the Chinese industry’s own account, already function as the default stack for developers outside the American ecosystem. Domestic chipmakers such as Huawei Ascend, Cambricon, Biren, now cover over 40 per cent of China’s own compute needs. By the time any plurilateral alternative gets its secretariat funded, a meaningful share of the infrastructure decision may already be made quietly, at the level of which repository an engineer chose to clone.

The involution precedent makes the point concrete. China has one word - 内卷, involution - for competition so wasteful it burns capital without building anything real. Regulators used it last year on the food-delivery war, where three platforms torched over 100 billion yuan in subsidies before Beijing ordered it stopped, and state media ran the verdict as a headline: “the delivery war should end”. That is state control exercised over a domestic industry in real time, and it is the same authority now steering capital into AI. Xi’s line about not “putting one’s own security above others’” reads differently once you know that.

A policymaker in Delhi finishing this piece should walk away with one fact: China’s economy needs the world to adopt its AI framework roughly as much as the world needs a functioning framework at all, and that mutual need is what lets Beijing lecture on governance from a position its own Q2 numbers show to be more fragile than the keynote let on. Whether the Global South accepts a Chinese-drafted architecture by default, or builds the plurilateral alternative before the standards lock in, is being decided now, by whichever government moves from declaration to delivery mechanism first. China already made that move. The clock started on July 17th, and nobody in the room asked Delhi’s permission to start it.

Lee Seung-ho did not borrow five times his savings because he wanted to become a trader. The 24-year-old student wanted an apartment before marriage. He turned 20 million won saved during compulsory military service into nearly 300 million won, then watched it disappear in four weeks as his brokerage sold his falling shares. Lee told Reuters that he would borrow again. His target remained a Seoul home, marriage and two children.

The distance between Lee’s savings and the life he imagines is written into Seoul’s housing market. The city’s median apartment price exceeded one billion won in 2024, having doubled in five years. In 2025, the mean age at first marriage reached 33.9 for men and 31.6 for women. The OECD estimates that the doubling of house prices between 2013 and 2019 reduced the likelihood of marriage by between 4 and 5.7 per cent. Where childbirth remains closely tied to marriage, the missing apartment delays more than the wedding. For an older generation, that distance was supposed to narrow through work. Korea’s government-led industrialisation created globally competitive exporters and attached a social promise to their success: education would lead to a secure company job, the job to property, and property to family life. Korea still builds advanced ships, cars, weapons and semiconductors. What has thinned is the route by which national prowess becomes ordinary security.

This ladder narrows at the entrance to stable employment. South Korea has the lowest share of large-company jobs in the OECD. The Korea Development Institute links this scarcity of “good jobs” to weaker social mobility, lower birth rates and the concentration of opportunity in Seoul. A few conglomerates offer wages and protections that much of the labour market cannot match. The economy remains world-class at the top while becoming less compelling below. Unable to close the gap through salaries, many young Koreans have turned to the companies that dominate both the economy and the share market. Samsung Electronics and SK Hynix now account for more than half of the KOSPI. Their gains rest on real demand for the high-bandwidth memory used in AI data centres. Yet the rally draws household savings into the family-controlled chaebol system, where founding families retain influence beyond their direct ownership. Recent corporate-law reforms aimed at strengthening smaller shareholders show that the old concentration of power remains unresolved. Borrowing has transformed participation into a wager. Margin-loan balances reached a record 38.63 trillion won in June, while total investor debt passed 60 trillion won in May. Regulators permitted funds that doubled the daily movement of individual shares, then blocked new listings two months later. The apartment market keeps the ants outside the property boom; the KOSPI brings them inside the corporate economy as leveraged minority shareholders. Work accumulates too slowly, housing recedes too quickly, and the trading app appears to compress the years between the two.

The same scarcity of believable growth shapes global markets. A collapse in Korean retail trading would hurt households, but it would not repeat 2008, when losses on mortgage-related assets spread strains through global financial markets. The greater danger lies in the story the ants have borrowed to join. Korea supplies the memory chips, American technology companies supply the data-centre spending, and investors everywhere price in the same hoped-for leap in productivity.

AI has become one of the few places where a slow-growing world can still imagine abundance. The IMF’s July 2026 outlook projects global growth of 3 per cent and finds that economies within the technology value chain are outperforming those outside it. Investors can retreat to economies offering modest earnings growth, or continue to push money towards the same narrow group of AI-linked firms. If data-centre spending slows, Korean chip orders and factory investment will weaken, followed by suppliers elsewhere in Asia and technology valuations in the United States. The collapse would not create stagnation. It would reveal how much momentum had been concentrated in one technological cycle.

The costs of Korea’s exhausted aspirational order already extend beyond markets. Delayed marriage is fuelling demographic contraction in a country where defence still depends on compulsory service. Between 2019 and 2025, the armed forces shrank by 20 per cent to about 450,000 personnel as the number of men reaching enlistment age fell. The defence ministry reported a shortfall of 50,000 personnel. President Lee Jae Myung is now pursuing selective conscription, a more technologically advanced military, and greater self-reliance within the American alliance.

Technology can reduce the number of soldiers required for specific tasks. It cannot detach military capacity from the society that supplies soldiers, workers and taxpayers. Korea’s corporations remain capable of extraordinary production, but the ordinary life that industrial growth was meant to secure has moved further away. The ants bet on AI because salaries no longer seem able to carry them towards it. When an apartment requires a jackpot, a booming market does not prove that aspiration survives. It shows that advancement has shifted from the wage packet to the wager.

“The right approach is to reveal enough to make the point, but not so much as to give everything away.” - Huang, Beijing-based scientist.

There is a number that used to be a point of pride for China, and now it’s being quietly walked back. It is the share of the world’s scientific papers written by Chinese researchers and published in journals the West built to grade itself, the Science Citation Index SCI. It is a ledger against which a scientist’s whole career is weighed. Twenty years ago, that share was 5 per cent. By 2023 it had climbed to 30 per cent. This was not an accident but a policy sustained over two decades, and it worked. Now in the middle of 2026, Beijing is discussing how to undo the very incentive that built it.

China’s total scientific publication output went from roughly 604,000 papers in 2018 to an estimated 1.31 million in 2025, more than doubling in seven years. The mechanism is dry and administrative, which is how you know it is serious. Chinese policy makers are weighing whether to strip the weight that SCI publication carries in decisions about academic promotion and tenure. For years, getting into Nature or Cell was a gateway to professorship and grants. Now several ministries, including the Ministry of Science and Technology, are drafting a framework that would fold national security considerations into how researchers get evaluated in the first place. If publishing abroad is what earns you a career, then publishing abroad is what you will keep doing, with or without a security review. Delink the reward from the act, and the leak risk shrinks on its own. The push sits inside a doctrine Beijing calls “zili gengsheng” or self-reliance, revived under Xi Jinping and folded into the 15th Five-Year Plan covering 2026 to 2030, which treats scientific independence as a matter of regime security rather than just economic policy.

The trigger for this round of tightening was a case the Ministry of State Security made public last month. A researcher, trying to improve the odds of getting a paper accepted at an international journal, had allegedly included what the ministry called important technical details that should never have left the building. In June, the security ministry went further, warning that submissions to international conferences, publications in foreign journals, and cross-border academic exchanges must all strictly follow rules requiring review before disclosure and approval before release. It cited another unnamed researcher who had reportedly folded “core equipment structures, key technical parameters and distinctive experimental sample data” into papers aimed at foreign journals and conferences.

None of this is happening in isolation. Denis Simon, a senior fellow at the Quincy Institute, put it this way: the Chinese science system has shifted from a catch-up mode to a great power mode. China no longer needs to catch up in the way it once did, in his telling. The question now is how it manages the knowledge it already has, protecting national security while still building its own scientific prestige.

There is also a less geopolitical reason for the reform, and the FT traces it back further than the current security push. In 2021, the human resources and education ministries had already warned universities against relying on “rigid” publication requirements, well before national security became the headline justification. The obsession with SCI publication, what Chinese commentators call SCI worship, has also fed academic misconduct: inflated publication records, papers built on manipulated or fabricated data. So the current move is not a clean pivot from openness to secrecy. It is an older complaint about research quality and a newer one about leaks, arriving at the same policy at roughly the same time.

Total Publications Per Year Between Countries

Base: all world SCOPUS recorded scholarly output published between the 1996 and 2022 calendar years

For the scientific community outside China, the practical effect is less about any single stolen formula and more about duplication. Researchers studying the trend warn that when major research powers withdraw into separate, security-screened tracks, the result is parallel vaccine pipelines, segregated AI models, and duplicated climate datasets, effort spent twice on problems that don’t respect borders. For Chinese researchers themselves, the immediate effect is narrower and more personal: a career ladder that spent thirty years rewarding one behaviour is now being told, without much clarity yet on what replaces it, to reward something else.

What the reform cannot easily fix is the alternative. Domestic Chinese journals, even the ones that make it into the SCI index, accounted for less than 5 per cent of the SCI output produced by Chinese researchers as of 2023. Telling a scientist to publish at home only works if home has somewhere to publish. The National Natural Science Foundation has already begun requiring that a fifth of the papers from its funded projects run in Chinese journals, a quota dressed up as a preference.

Not every Chinese scientist sees the retreat as a loss. But there is the quieter tension sitting underneath the security framing. A system built for thirty years to chase prestige in someone else’s journals cannot be redirected by ministerial memo alone. The prestige has to go somewhere, and right now nobody in Beijing has quite decided where.

Sri Lanka returned to the ranks of upper-middle-income countries by thirty-four dollars. The figure carries the authority of a verdict and the slight absurdity of an accounting error. Under the World Bank classification that took effect on 1 July 2026, Sri Lanka’s gross national income per capita for 2025 was $4,670. The threshold was $4,636. No economy feels transformed by an additional thirty-four dollars, but statistical categories do. Four years after defaulting on its foreign debt, rationing petrol and watching its president flee the country, Sri Lanka found itself in another column of the global ledger.

It is a return rather than a first ascent. Sri Lanka had entered the same category in 2019 and slipped back the following year. The reclassification captures a genuine recovery in income, growth and institutional control. It contains no memory of the nights spent in fuel queues, of pharmacies without medicine, or of children arranging schoolwork around electricity cuts. National income had travelled far enough to cross a line. The people gathered inside the average had not all travelled the same distance. At the height of the shortages, the Sri Lankan state appeared on a telephone screen. The National Fuel Pass assigned each registered vehicle a QR code and a weekly petrol allocation. It created no fuel. It imposed sequence upon scarcity: this vehicle before that one, this many litres and no more. For a few months, one of the state’s most intimate functions was to decide how far a citizen could travel.

Seventy years earlier, scarcity had been administered through another object small enough to fit in the hand. The rice ration book carried the assurance that the staple around which domestic life revolved would remain within reach. When the government increased the price of rationed rice from 25 to 70 cents a measure in 1953, trains stopped, roads were blocked, and workers stayed away. The Hartal of 12 August turned the price of a meal into a judgment on the meaning of independence. The ration book and the QR code belonged to different political worlds. The first emerged from a state that extended its authority through welfare; the second from one that tried to preserve its authority after bankruptcy. Both reveal the intimacy of Sri Lanka’s economic history. A shortage rarely remained inside the Treasury or the Central Bank. It moved into the rice pot, the schoolroom, the bus journey and, eventually, political allegiance.

Sri Lanka’s postcolonial state had made a social promise larger than its modest income appeared able to support. Free education, public healthcare, food subsidies and social services were not rewards postponed until industrial prosperity arrived. They were intended to create a more equal society before the country became rich. The achievement unsettled the assumption that growth must precede welfare. A child born on a relatively poor island could expect schooling, medical care and a life considerably longer than income alone predicted.

Laksiri Jayasuriya described this settlement as a form of social citizenship. A public hospital did more than deliver a service; it instructed citizens in what government was for. Subsidised rice placed the state at the family meal. Free education opened routes of social mobility, although language policy, regional inequality and the disenfranchisement of plantation Tamils revealed how unevenly citizenship had been distributed. Welfare widened belonging while the same state drew exclusions around it.

Later accounts sometimes treated this arrangement as Sri Lanka’s original economic sin: a poor country promising too much, then borrowing for decades to avoid admitting that it could not afford the promise. Schools and hospitals did not bankrupt the island. The deeper weakness lay in the productive and fiscal structure beneath them. Sri Lanka became more successful at distributing the social gains of development than at transforming the economy required to sustain them. The island remained exposed to the world beyond it. Colonial Ceylon earned foreign exchange through tea, rubber and coconut. Garments, tourism and migrant labour later broadened those earnings without ending their vulnerability. Fuel, medicine, machinery, fertiliser and many inputs needed for domestic production still had to be purchased in foreign currency. The economy could appear settled in rupees while growing anxious in dollars. This gave Sri Lankan politics a distinctive relationship with time. When foreign currency became scarce, governments could impose changes with immediate costs or wait for the next inflow. They became highly practised at waiting. A weak export year might be repaired by a better tourist season. Remittances could compensate for jobs that the domestic economy had not created. A loss-making public enterprise could survive because reform would disturb workers, consumers and political appointments at once. Borrowing allowed each problem to be advanced without requiring the entire structure to be confronted. Financial rot accumulated through these comprehensible postponements. It was not a secret illness discovered in 2022. It was the public habit of refusing to make several weaknesses visible simultaneously.

Saman Kelegama’s account of post-independence development approached the economy from this side of the dilemma. He saw recurring foreign-exchange constraints, limited productive transformation, electoral pressures and conflict. Read beside Jayasuriya, his argument removes the false choice between celebrating welfare and blaming it. Sri Lanka’s social settlement deserved preservation. The economy beneath it remained perilously shallow.

The opening of the economy in 1977 changed how that contradiction was managed. Imports returned to shop windows, garment factories expanded, tourists arrived, and migration connected rural homes to wages earned in the Gulf. J. R. Jayewardene offered consumption as proof that scarcity had ended, while the new executive presidency concentrated the authority through which the open economy would be governed. Markets widened, yet the state did not disappear. Public employment, subsidies, and state enterprises survived alongside contracts, licences, and opportunities distributed through political proximity.

The civil war hardened this political form. It enlarged the military, normalised emergency rule and secrecy, and accustomed the executive to treating scrutiny as an obstacle to national survival. Its costs were never merely fiscal. The north and east endured destruction, displacement and the erosion of livelihoods; elsewhere, the war allowed national indicators to coexist with sharply different regional economies. As analyses of Sri Lanka’s hybrid political regime have shown, the conflict did not mechanically cause the default thirteen years after it ended. It shaped the state that would later meet economic danger: centralised, securitised and impatient with disagreement.

When the war ended in 2009, the Rajapaksa government did not simply claim a peace dividend. It transferred the language of victory into development. Roads, ports, airports and a remade Colombo would bind the island materially after the state had defeated the challenge to its territorial authority. Mahinda Rajapaksa’s political character suited the moment. He spoke with rural familiarity while presiding over concentrated power; the family appeared less as temporary occupants of office than as custodians of the national future.

Concrete made that claim visible. Infrastructure had genuine uses, and Chinese lending alone cannot explain the bankruptcy. The more consequential shift was Sri Lanka’s growing reliance on international sovereign bonds and other commercial debt, which arrived quickly and demanded repayment on shorter, less forgiving terms than concessional loans—borrowing compressed political time. A port could rise before the economy around it existed; the prestige of development arrived immediately, while repayment belonged to a future government. Saman Kelegama examined the politics and contradictions of this period in his study of the Rajapaksa decade.

Gotabaya Rajapaksa inherited the family mythology but not his brother’s political elasticity. Elected in 2019 after the Easter bombings, he offered the severe competence of a former defence official. His preference for command over negotiation shaped the mistakes that followed. According to the IMF’s 2022 assessment of Sri Lanka, government revenue and grants fell from 12.6 per cent of GDP in 2019 to 9.2 per cent in 2020, while the fiscal deficit widened from 8 to 12.8 per cent. Tourism had already weakened; the pandemic then emptied hotels and shut Sri Lanka out of international capital markets. The government attempted to preserve stability by defending the exchange rate, restricting imports and financing spending through the central bank. The fertiliser ban of 2021 revealed the same governing instinct. A difficult agricultural transition was declared as an immediate national transformation, with farmers expected to reorganise cultivation at the speed of an executive announcement. The larger crisis was already taking shape. Sri Lanka was losing the dollars needed to sustain an import-dependent economy, and its rulers were losing credible means of obtaining them.

Debt service continued while reserves disappeared. Import controls concealed a shortage of foreign currency by converting it into shortages of goods. In April 2022, the government suspended external debt payments. By the end of that year, usable reserves had fallen to about $500 million, according to the World Bank’s April 2026 Macro Poverty Outlook. Adjustment had not been avoided; it had been delayed until it could occur only through collapse. By September, Colombo inflation had reached 69.8 per cent and food inflation 94.9 per cent. Inflation at that scale changes the character of time inside a household. A salary ceases to be a monthly quantity and becomes something that loses meaning each day it remains unspent. Savings accumulated over the years can disappear if not used; the future contracts to the next purchase.

Families became the shock absorbers of the economy. Meals were simplified, medicines were rationed, and purchases were postponed. Those with dollars or relatives abroad inhabited depreciation differently from those dependent on fixed rupee wages. Women carried much of the unpaid work created when transport, public services and consumption failed together; the country’s main foreign-exchange earners in garments, plantations and migrant labour already depended heavily on their paid labour. As Kanchana Ruwanpura, Bhumika Muchhala and Smriti Rao have argued, the debt crisis was also gendered. Remittances appeared as resilience in the balance of payments, but they were wages earned through absence, and households reorganised across borders.

The Aragalaya grew out of this exhaustion, though it was never solely a protest against prices. “Gota go home” gave a long crisis a name because political movements require a scapegoat. When protesters entered the presidential residence, the contrast between public scarcity and private abundance briefly occupied the same rooms. The movement removed a president and broke the authority of a family that had made itself difficult to distinguish from the nation. Gotabaya Rajapaksa fled the country in July 2022, but the movement could not restructure a bond.

Ranil Wickremesinghe represented another political temperament. He belonged to the old Colombo establishment, spoke the language of institutions and international finance, and possessed little popular intimacy. Parliament elected him with support from the formation the streets had rejected. His government restricted protest, but it also accepted the facts the previous administration had resisted: the debt could not be paid as contracted, the currency could not be permanently defended, and the state could not recover without collecting more revenue.

The $3 billion IMF programme approved in March 2023 formalised that recognition. Debt was restructured; taxation increased; fuel and electricity prices moved closer to cost; monetary financing was curtailed; the exchange rate became more flexible; and the central bank gained greater autonomy. These were not rituals performed solely for creditors. A state without reserves cannot import medicine through goodwill, and inflation punishes households whose incomes cannot be renegotiated. Necessity did not determine distribution. Indirect taxes reached into consumption. Electricity tariffs moved losses from public accounts into household and business bills. Currency depreciation compressed imports because people could afford fewer of them. The government balance sheet and the family budget were repaired through the same process. Costs earlier borrowed, subsidised, or concealed reappeared in higher tax payments, higher prices, and diminished consumption.

Aswesuma gave this relationship an administrative form. The new welfare programme sorted applicants into transitional, vulnerable, poor and severely poor groups. Targeting promised to reduce the patronage and leakage associated with Samurdhi. It also required deprivation to become legible before it produced an entitlement. The World Bank’s Public Finance Review later found that district allocations still relied on 2019 poverty patterns, although the crisis had altered where insecurity lived. A family could become poor faster than the database could learn its new condition. The recovery was nevertheless real. By the end of 2025, usable reserves had risen from roughly $500 million to $5.4 billion, enough for about three months of imports. The economy had grown by 5 per cent for a second year; the current account recorded a $1.7 billion surplus. Government revenue and grants reached 16.7 per cent of GDP, and the primary balance moved to a surplus of 5.4 per cent. These figures, recorded in the World Bank’s April 2026 economic assessment, described a state that had regained substantial control over itself. Fuel queues disappeared, medicine returned, and firms could plan purchases without wondering whether the rupee would lose another portion of its value before a shipment arrived. The figures also contain quieter stories. The unusually large primary surplus reflected strong tax collection, especially after vehicle imports resumed, and the under-execution of capital spending. Reserve accumulation rested on tourism and remittances, including the labour of citizens who had left. As the IMF’s 2026 review makes clear, recovery restored circulation: dollars entered, imports arrived, taxes were collected, and obligations were renegotiated. It did not yet answer whether the economy producing those flows had changed.

Anura Kumara Dissanayake’s election in 2024 promised a political break from dynasties and the established parties. His National People’s Power coalition won 159 of 225 parliamentary seats. Yet his government retained the central framework of the IMF programme. Bankruptcy had narrowed the range of credible rupture. A new administration could alter priorities, expand protection and confront corruption; it could not behave as though default had never happened. Wickremesinghe imposed discipline without a popular mandate. Dissanayake inherited a mandate for change within limits drawn by creditors and by the memory of empty filling stations.

This constrained recovery met its first major test not in a negotiation room or a bond market, but in rain. Cyclone Ditwah made landfall on 28 November 2025, bringing prolonged rainfall, floods and nearly two thousand landslides. By 19 January, 646 people had died, and 173 remained missing. Close to two million people were affected across the island, while homes, farms, roads, railways and water systems were damaged. The storm arrived just as the figures of recovery appeared most persuasive. The World Bank’s rapid assessment placed direct physical damage at $4.1 billion, approximately 4 per cent of GDP. That figure could record a destroyed bridge or flooded house, but not the wages lost while roads remained impassable, the debt assumed to repair a roof, or the cultivation season that could not be recovered. Ditwah did not cause Sri Lanka’s bankruptcy. It changed the meaning of recovery. Reserves and tax revenues had improved just as reconstruction created new fiscal and foreign-exchange demands.

Sri Lanka therefore returned to the IMF while the existing Extended Fund Facility remained in place. In December 2025, the Fund approved approximately $206 million through its Rapid Financing Instrument. The contradiction was immediate: a state that had created fiscal room through higher taxation, restrained spending and the under-execution of capital budgets now required precisely the public investment that consolidation had limited. The burden also fell upon households that had not recovered from the original crisis. In 2025, 22.1 per cent of Sri Lankans remained below the World Bank’s $4.20-a-day poverty line, while 32.8 per cent of households experienced moderate or severe food insecurity. The Middle East conflict poses a secondary danger by raising commodity prices, disrupting essential supplies, and weakening tourism and migrant earnings. Ditwah was the immediate test. Sri Lanka became governable again, but the cyclone asks whether a state trained by bankruptcy to tax, target and restrain can also rebuild, protect and invest without requiring vulnerable households to absorb another national emergency.

Sri Lanka did not travel from bankruptcy to prosperity in four years. It became governable again. That achievement matters to anyone who remembers the queues, because predictable prices, available fuel, and a currency able to purchase medicine are forms of security. Yet governability is not a renewed social compact. The state has recovered its ability to measure, tax, target and restrain. It has not shown that these capacities will create an economy less dependent on borrowed time and the departure of its workers.

There is a number that the Chinese state publishes every month, and another that it publishes with more reluctance, and if you lay the two beside each other on a desk they begin to tell a story that neither was meant to tell alone. The first is the concentration of fine particulate matter in the air, the soot and sulphates and nitrates small enough to enter the blood, measured in micrograms per cubic metre. The second is the share of young people between sixteen and twenty-four who are looking for work and cannot find it. One is a measure of how much the country is burning. The other is a measure of how little the burning now buys or sells at home.

For a decade the first number fell. This was one of the genuine achievements of the age, and it is worth pausing on before the story turns, because the story does change. In 2013 the air over China’s cities was a byword for catastrophe. By 2019 the average concentration of PM2.5 had dropped by half. Factories were fitted with scrubbers, coal plants were closed or moved, the north was heated by gas instead of briquettes. A person who had left Beijing in 2013 and returned in 2020 would have looked up and seen a different sky. The state had promised blue skies and, for once, more or less delivered them.

Then the pandemic ended, and the number stopped falling. In 2023 it rose for the first time since the campaign began. The Ministry of Ecology and Environment said, with a candour that is rare and therefore instructive, that the rebound was tied to a surge in energy-intensive industrial production after the restrictions lifted. When a country wants to grow again in a hurry, it makes steel and cement and chemicals, and when it makes those things the air tells on it. The sky is an honest accountant. It records what the factories do regardless of what the ministries say.

By 2025 the national average had settled at around 28 micrograms per cubic metre. This is within China’s own standard of 35. It is also roughly six times what the World Health Organisation considers safe. Both of these facts are true and the gap between them is the whole matter. The country has cleaned its air enough to satisfy itself and not nearly enough to satisfy the lungs of the people breathing it. And the improvement, such as it is, has grown uneven in a way that maps almost exactly onto where the economy is straining. In the first quarter of 2025 the national figure fell by five per cent, which sounds like progress, until you look at the provinces. In Guangxi the concentration rose by thirty-two per cent. In Yunnan by fourteen. In Xinjiang by eight. These were not the accidents of a bad wind. They were the residue of industry moving west, chasing cheaper land and looser enforcement, carrying its smoke with it.

Now set the second number down.

In the summer of 2023 the share of urban young people out of work reached twenty-one per cent, a record, and the state did a curious thing. It stopped publishing the figure. For six months there was silence, and then the number returned in 2024 wearing new clothes, recalculated to exclude students, which lowered it. Even in its gentler form it has stayed high. In August of 2025 a record twelve million university students graduated, and the youth jobless rate climbed to nearly nineteen per cent. This spring it sat near seventeen. In May of 2026 it eased to fifteen and a half, the lowest in eleven months, which the ministries reported as good news and which is good news the way 28 micrograms is clean air, that is, only against a standard that has already made its peace with the trouble.

Here is where the two numbers speak to each other. The economy is burning enough coal to dirty the sky over Guangxi and Yunnan. The factories are running. The chemical plants and the steelworks and the thermal power stations are lit. And yet the young cannot find work in them, because the work has changed. The factories that stayed open increasingly run on machines. The twelve most labour-intensive industries in the country shed three and a half million jobs between 2019 and 2023. The smoke rises but fewer hands are needed to make it rise. This is the thing the pollution data quietly reveals that the growth figures conceal. Industrial activity and industrial employment have come apart. The country can pollute at full capacity and still leave a fifth of its graduates idle, because the modern factory is a place of few workers and many robots, and the robot does not job-hunt and does not appear in the unemployment line.

There was a third number once, though nobody wrote it down, and it was the hours. For the better part of two decades the bargain in the Chinese city was plain enough that people gave it a name. They called it 996. You worked from nine in the morning to nine at night, six days a week, and in exchange you were paid and you belonged to the machine that was lifting the country. Jack Ma of Alibaba called the arrangement a blessing in 2019 and said that if you wanted to join his company you had better be ready to give it twelve hours a day, and why else would you bother. Richard Liu of JD.com wrote that slackers were not his brothers. These men were stating the terms of the pact aloud, and the terms were these: your youth and your health and your evenings, in return for a place on the escalator that was carrying everyone up. For a long time the young signed. The escalator was real and it was moving and the price, however steep, bought a ticket that seemed worth having.

Then in March of 2019 a programmer opened a page on GitHub and called it 996.ICU, the letters standing for the intensive care unit where the schedule would eventually land you, and something broke open. Workers began posting the names of the companies that demanded the hours. They voted. They compared. The browsers made by Tencent and Alibaba and Xiaomi quietly blocked the page, which told you the nerve had been touched. This was not a strike in the old sense. It was quieter and stranger, a refusal conducted in code and comment threads, but it was the first time the young had said the price was too high in a voice the whole country could hear. The Supreme People’s Court ruled the schedule illegal in 2021. Enforcement wandered. Companies renamed the thing “big week, small week” and carried on. In 2024 a Baidu communications chief said it was not her business if her employees’ personal lives were destroyed by their work, that she was not their mother, and she resigned within days under the weight of the reaction. The bargain was cracking not because the state decreed it but because the people inside it had stopped believing the escalator was still moving.

And this, finally, is what the sky over Chengdu is telling us. A superpower makes a pact with its own people, and the pact is growth. You give me your labour and your lungs and your long hours, says the state, and I will give you a life better than your parents had. For thirty years both sides kept the terms. The air darkened and the wages rose and the cities filled with people who had come from villages and would not go back. The smoke was the visible sign of the transaction, hanging over Beijing and Tianjin, and the young breathed it because the breathing bought something. That was the devil’s part of the bargain, and the devil was paid on time.

What happens when the pact changes is what we are watching now. The state still asks for the labour, but the growth it offers in return has thinned. The factories still burn, but they no longer hire. The young are still told to sacrifice, but the escalator has slowed to a crawl and some of them can no longer see where it leads. A generation raised on the promise that hard work and a degree would carry them upward has arrived to find the smoke still rising and the jobs gone into the machines, and they have begun, quietly, in code and in silence and in the flat refusal that the Chinese have named “lying flat,” to decline the terms their parents accepted. When a nuclear firm posts eight thousand jobs and one million people apply, the young are not lazy. They are looking at a bargain that has stopped paying and deciding, some of them, not to sign.

The delivery driver on his scooter, degree in one pocket and phone in the other, threading through the traffic of a city whose air is a little cleaner than it was and a little dirtier than last year, is the figure in whom all of this meets. He is the labour the pact still wants and the prosperity it can no longer promise. He works hours his parents would recognise and earns a future they would not. No state can hold a bargain together forever when only one side is still being paid, and the deepest question hanging over China now is not whether the sky will clear or the economy will grow. It is what a country does when the generation it was built to reward looks up at the smoke, does the arithmetic, and concludes that the old exchange of breath for bread is a trade it would rather not make. The devil, in these arrangements, is patient. He waits to be paid. And when the young decide the debt is no longer theirs, someone else must settle it, and no one yet knows who.