The clean version of the Iran story is that negotiations are moving, sanctions are tightening, oil traders are guessing, and Washington is trying to move a war toward a settlement. The truer version is uglier and more useful: everyone is counting inventory.

The United States imposed new sanctions on an Iranian agency accused of trying to control shipping in the Strait of Hormuz, the waterway through which about a fifth of the world's oil and natural gas normally passes. The measure is meant to add economic pressure on top of military force while the Trump administration says a deal is near and talks continue. On Wednesday, Reuters reported oil settled 5% lower as traders priced in hope for a framework that would reopen Hormuz. By Thursday in Asia, AP reported world shares were mostly lower and oil prices were up more than $2 after fresh US strikes against Iran.

That is the visible market. The more consequential market is the one without a public ticker: Patriot interceptors, THAAD missiles, shipping capacity, refinery margins, political patience.

Ukraine made the shortage explicit. President Volodymyr Zelenskyy confirmed that he had written President Trump and Congress asking for more US-made air defense ammunition, including Patriot PAC-3 missiles, as Russia intensifies missile attacks. The plea comes as the Iran war diverts American stocks. Axios, citing a Center for Strategic and International Studies analysis, reported that US use of air-defense interceptors in the Middle East has created a gap likely to last years, with replenishment not expected until 2029 under current assumptions. The same report said the United States received 172 Patriot interceptors in fiscal 2026 and has used more than 1,000 in the Iran war.

This is how wars leak into one another. A missile fired in the Gulf becomes a battery left thin outside Kyiv. A tanker moving through Hormuz becomes a gallon of gasoline in Denver. A sanction announced in Washington becomes a price on a screen in Seoul. The map is less a set of fronts than a single supply chain with flags on it.

Trump's political problem is that he wants the appearance of closure before the mechanics of closure exist. A deal that says Hormuz will reopen is not the same as insurance underwriters accepting the route, tankers transiting freely, mines being cleared, oil cargoes moving at scale, prices staying down, and allied missile cupboards no longer emptying faster than factories can fill them. The distinction matters because American voters do not experience foreign policy as communiques. They experience it as the cost of driving, the shakiness of markets, and the uneasy knowledge that an ally under Russian attack is asking for weapons the United States has been spending somewhere else.

Iran also understands inventory. The more it can make reopening conditional, partial, slow or reversible, the more it can turn geography into leverage. Washington understands it too, which is why sanctions are aimed not only at punishment but at who controls the chokepoint's practical administration.

The war is therefore not ending at the podium. It is ending, if it ends, in shipping manifests, production lines, port risk assessments and procurement contracts. The front page word is not peace. It is capacity.

Ukraine's latest request to Washington is not an abstract plea for support. It is a specific request for interceptors, made after Russian strikes showed again that ballistic missiles remain Moscow's most punishing advantage.

Zelenskyy told Trump and Congress that Ukraine needs more Patriot PAC-3 missiles and other air-defense systems. AP reported that the Ukrainian leader warned deliveries are falling dangerously short as the Iran war diverts US stocks. The timing is brutal. Russia used a hypersonic Oreshnik ballistic missile during a mass attack on Kyiv on Sunday, according to Ukrainian officials, part of an assault that included hundreds of drones and scores of missiles.

For Kyiv, the shortage is not merely military. It is civic. Every unloaded battery changes the calculation for schools, hospitals, train stations, apartment blocks and government offices. Air defense is the difference between a city that can function under attack and a city forced to live by siren.

The United States has insisted it retains enough capability for its own missions. That may be true. It does not answer the allied question: what happens when the same finite industrial base is asked to cover the Gulf, Ukraine, Taiwan planning and homeland defense all at once? In the old language of strategy, alliances were promises. In the present language, they are production schedules.

The Iran war has exposed a hard fact that procurement briefings often soften. High-end defensive missiles are exquisite, scarce and slow to replace. Russia can exploit that asymmetry by launching salvos that force Ukraine to choose what to protect. Iran can exploit it by forcing the United States to spend advanced interceptors at a rate factories cannot match.

Ukraine is not asking America to notice a new war. It is asking America not to let one war cannibalize another.

The climate story keeps being written as a warning, which is comforting because warnings imply time. The latest UN-linked projection reads more like a calendar.

AP reported that the World Meteorological Organization and the United Kingdom's Meteorological Office now see a 75% chance that the 2026-2030 average global temperature will exceed 1.5 degrees Celsius above pre-industrial levels. The forecast also gives a 91% chance that at least one of the next five years will cross that mark, and an 86% chance that one year will break the hottest-year record set in 2024.

This does not mean the Paris threshold has formally failed in the legal sense, because the long-term benchmark is measured over decades. It does mean the lived world is increasingly operating inside the conditions diplomats once described as a line not to cross.

Europe supplied the immediate evidence. The United Kingdom broke a century-old May temperature record twice in 24 hours, with 35.1 Celsius recorded at London's Kew Gardens. France also saw severe heat, and deaths were reported as people tried to cool down. The striking detail is not simply the number. It is the timing. May is no longer reliably a preface to summer; it can now behave like a rehearsal for the dangerous part.

Climate politics often gets trapped between apocalypse and delay. The useful middle is operations. Heat plans, power grids, worker protections, urban cooling, fire staffing and water rules are no longer adaptation theater. They are the basic machinery of government.

The next five years are not a prophecy. They are a work order.

The week's immigration cases have a common question underneath the legal variety: who gets heard before the machinery moves?

A federal appeals court gave Mahmoud Khalil, a former Columbia University graduate student and lawful permanent resident, more time to fight the Trump administration's effort to deport him. Khalil was detained last year over his role in pro-Palestinian demonstrations. AP reported that the 3rd Circuit said his case should proceed first through immigration courts, but put its ruling on hold while he appeals to the Supreme Court.

In another case, the Supreme Court sided with the Trump administration in a dispute over speech restrictions for immigration judges. The justices said the judges' challenge belongs in the federal employee complaint system, not ordinary federal court. Immigration judges occupy an awkward status: they decide cases, but are federal employees inside the executive branch rather than Article III judges.

Then came the Kilmar Abrego Garcia ruling. A federal judge dismissed human smuggling charges against Abrego Garcia, finding the prosecution was designed to punish him for challenging his mistaken deportation to El Salvador. The judge called the case an abuse of prosecuting power.

None of these rulings settles the whole immigration fight. Together, they mark the terrain. The administration wants speed, central control and limited outside review. Its challengers want federal courts to remain open before removal, detention or retaliation become irreversible.

Procedure can sound bloodless. In immigration law, it is often the last available defense against a one-way door.

The economy is currently telling two stories, and only one of them can afford a victory lap.

AP reported that US consumer confidence slipped this month while gas prices stayed high and inflation remained elevated, even as stock prices hovered near records. A survey found that two out of three Americans are cutting back on spending. That is not a contradiction. It is the distribution table.

The asset-owning economy is being pulled upward by AI, megacap earnings and hopes that the Iran shock can be contained. The household economy is getting revised at the pump. Reuters reported Wednesday that oil settled sharply lower on hopes for US-Iran progress and a reopening of Hormuz, but AP reported Thursday that fresh US strikes helped push oil prices back up in early global trading. Traders can change their minds in minutes. Gasoline budgets cannot.

The Federal Reserve is trapped in the same split-screen. Minutes from the April 28-29 meeting show the central bank left its rate range unchanged and kept the primary credit rate at 3.75%. That stance fits an economy where inflation has not fully behaved and energy is still a live shock.

The reader question worth keeping is whether oil matters less in an AI economy. It matters differently. Data centers can move the S&P 500. Gasoline moves the household mood. The first gets modeled by analysts; the second gets felt before work, before school, before every long weekend trip that suddenly looks optional.

Markets are allowed to be forward-looking. Voters are allowed to be receipt-looking.

Nvidia's latest quarter did something clarifying. It made the AI boom too large to describe as vibes.

The company reported record fiscal first-quarter revenue of $81.6 billion, up 85% from a year earlier, with data center revenue of $75.2 billion, up 92%. It also authorized an additional $80 billion in share repurchases and raised its quarterly dividend from one cent to 25 cents per share. Those are not startup numbers. They are infrastructure-rent numbers.

The scale matters because AI is increasingly a capital allocation contest. Models get the headlines, but access to chips, power, cloud contracts and secure deployment channels determines who can actually compete. Reuters reported this month that Microsoft, Google and xAI agreed to give the US government early access to new models for national security testing, adding to earlier arrangements involving OpenAI and Anthropic. The frontier model business is now entangled with procurement, standards work and military anxiety.

At the same time, Reuters reported that Anthropic committed to spend $200 billion with Google Cloud and chips over five years, according to The Information. Even allowing for the caution appropriate to reported contract figures, the direction is unmistakable: the AI economy is consolidating around firms that can prepay for scarcity.

The public still talks about chatbots as products. The companies are behaving as if the product is the industrial base itself. Nvidia sits at the tollgate, counting.

The culture ledger had two winners this week, and neither got exactly what it wanted.

At Cannes, Cristian Mungiu's Norway-set drama 'Fjord' won the Palme d'Or, making the Romanian filmmaker one of the rare directors to take the festival's top prize twice. AP described a festival where few films caused a large stir, but Mungiu's drama about polarization found broad admiration. The film's premise is engineered for argument: Romanian Evangelicals move to Norway and collide with state power, child welfare and competing moral languages.

At the box office, 'The Mandalorian and Grogu' opened with $82 million in the US and Canada and was expected to reach $102 million domestically by the end of Memorial Day. That is healthy. It is also low by Disney-era Star Wars standards, closer to the four-day Memorial Day frame of 'Solo' than to the old imperial dominance of the brand. The difference is cost, expectation and streaming context: this movie emerged from television and was made for less than the most bloated franchise chapters.

The split is instructive. Cannes rewarded seriousness about social fracture. The multiplex rewarded comfort with a known universe, but not with blank-check fervor. Audiences have not abandoned scale. They have become more discriminating about what kind of scale deserves theatrical urgency.

Taste is doing what politics often cannot: admitting that consensus is gone, then asking what is still worth gathering for in a room.

The most abused word in foreign policy is not victory. It is ceasefire.

A ceasefire should mean that the killing pauses and the system begins to relax. In the Iran conflict, the word has too often functioned as upholstery for a continuing crisis. Ships are not yet moving normally. Oil is still trading on rumor, retaliation and risk. Sanctions are being added while talks are supposedly progressing. Fresh strikes still move markets. Ukraine is asking for air-defense missiles because American stocks have been consumed elsewhere.

This is not pacifist fastidiousness. It is accounting. If a war declared to be ending is still raising gasoline prices, constraining allied defense, shaping midterm politics and determining what tankers can insure, then the war remains active in every arena that matters.

The temptation for governments is obvious. The past tense is politically useful. It lets leaders claim resolve and restraint at once. It allows them to tell hawks that force worked and voters that relief is coming. It makes ambiguity sound like strategy.

But readers should be stingy with the word peace. A ceasefire that depends on a chokepoint, a sanctions list, an ammunition shortage and a mood swing in oil futures is not yet peace. It is an argument with fewer cameras.

The test is simple: when the war's costs stop arriving under other names, then call it ending.

The public AI debate still pretends the main question is whether Washington will regulate. Washington is regulating already. It is doing so through contracts.

The White House has urged Congress to preempt state AI laws and favor a light federal framework. That is the visible ideology: innovation, uniformity, fewer local constraints. But the operating state is more muscular. Reuters reported that Microsoft, Google and xAI agreed to give the US government early access to new models for security testing, building on prior arrangements with OpenAI and Anthropic. The Pentagon has also been striking deals to put advanced AI capabilities on classified networks, while disputes over acceptable military use have become supply-chain questions.

This is how policy often arrives in America: not first as a statute, but as a procurement condition. A company wants access to government markets, classified deployments, cloud contracts or official confidence. The government wants access to model weights, evaluations, threat data and veto points. Both sides call it partnership because partnership sounds less alarming than dependency.

Some of this is necessary. Frontier models may create real cyber and security risks. Government cannot responsibly wait for a perfect omnibus law. But contracts are a narrow form of democracy. They bind the firms at the table, privilege insiders who can afford compliance, and leave the public reading press releases after the architecture has hardened.

The AI safety regime is not waiting to be born. It is being signed, clause by clause, by people who will later tell Congress what is already practical.