Thirty minutes later, Iran’s Foreign Ministry said the signing would not happen tomorrow. They said it… well, could happen “in the coming days.”

If you watched my latest video, you know where this is going.

The pattern I described then: Trump announcing a deal, the market pricing it in, and the terms getting disputed in public before anyone signs anything, ran again this week, in fast-forward, and now with a date stamp. The S&P added over a trillion dollars in market value on Wednesday’s announcement. Oil collapsed below $87. All of that happened before a single signature.

Now, I want to be fair here, because there are two ways to read what’s happening. The first is that Trump is genuinely close to something real, and the public contradictions between Washington and Tehran are what negotiations look like when they’re happening across a dozen countries and two governments that don’t trust each other. Pakistan’s prime minister said a deal was “closer than ever before” and expected to be finalized within 24 hours. That’s three sovereign governments triangulating toward the same word: tomorrow.

The second reading is that Trump needed this signed before Sunday, because Sunday is his 80th birthday, and a peace deal on your 80th birthday is the kind of thing you mention at every rally for the next six months before any midterm. And also, a very, very Trump thing to do.

I honestly don’t know which one is true. Maybe both. The behavioral read is focused on the gap between what someone says and what they do, and right now the gap between “scheduled to be signed tomorrow” and “we have not yet made a final decision” is the width of an ocean.

But here’s what I do know: The Islamabad Memorandum, named after Pakistan’s capital, not coincidentally, since Pakistan is the country that made the three phone calls that stopped the bombers three hours before they flew, calls for the Strait of Hormuz to reopen, the naval blockade lifted within 30 days, a 60-day ceasefire extension including Lebanon, and Iran committing to never acquiring a nuclear weapon. That last line is what Trump keeps leading with.

Sounds… clean? Sounds like a win, to be honest.

Any actual action on Iran’s nuclear program: the physical hardware, the enriched material, the centrifuges, only happens if a second deal is reached. A separate negotiation. To be held during a 60-day window that starts after the MOU is signed. The 2015 nuclear agreement, which everyone agrees was incomplete, took roughly 18 months to negotiate. So, sixty days is a bit of wishful thinking.

And Iran’s foreign minister said, on the record, that his preferred method for dealing with the enriched uranium is to dilute it inside Iran. Under inspectors, yes. Inside Iran, yes. Which means the material stays in the country while the second negotiation runs its clock.

When Trump says this is a “wall to no nuclear weapon,” he is describing the commitment Iran is making about the outcome. He is not describing what happens to the 440 kilograms of enriched material that produces the weapon. Those are two different sentences, and only one of them is in this MOU.

To be clear: this might still be the best deal available, given everything I described on my video about why you can’t simply bomb the country flat and call it done. The constraints are real. The GCC’s preferences are real. But “best available option under the circumstances” and “the deal Trump described tonight on Truth Social” are not always the same document, and understanding the gap between those two things is exactly why you’re here.

There are two other stories from this week that connect to all of the above in ways nobody is putting together yet, and one of them involves a number that makes Iran’s $300 billion demand look like a rounding error:

Read more

Doesn’t America have the guns to thoroughly glass Tehran three times over? Of course it has the firepower. Only fools could question that fact. It demonstrated it has the ability to surgically remove dictators at will, either in a Nike jumpsuit or a bodybag, whichever comes first.

Then why can’t Trump just ‘finish the job’ and stop what has been dragging for 100 days?

Ok, let’s do one of my favorite intellectual exercises and imagine that something Michael-Bay worthy happens. Trump does the nod, Hegseth nods in return, and Tehran, as well as any relevant IRGC outpost, bridge and power plant in Iran is turned into a Walmart parking lot. Just with a little more rubble than an actual Walmart parking lot.

Got that image in your head? Nice. What do you expect the Iranian regime is going to do in that situation? They will still have several dozen missiles and a couple hundred drones to launch. To... launch at what? Well, if I had to make a guess, I’d say they’d launch several of those at the UAE, or Bahrain, or Kuwait. Especially the latter two, since the UAE has demonstrated robust defense systems and the IRGC would want to do one and only one thing:

Maximum damage to whoever in the GCC that has even remote ties to Washington. Or despises the Iranian ‘ruling model’ of the past 47 years. Or both.

That’s the problem when you look at this from a ‘burn the ships’ perspective. Of course you’re not burning your own ships in this case, you’re burning the Iranian regime’s leeway to... well, exist as a proper country. Especially the ‘bomb the bridges and power plants’ part. That, in fact, will affect innocent Iranians more than the regime itself. In fact, that would actually turn neutral Iranians AGAINST America as well.

But the main problem for the Trump administration would be the *cornered* Iranian regime. Because we all know what happens when you corner rats. Or regimes, in this case. Burned ships mean there is no possibility of retreat and all that is left is to fight to the death and take with you as many enemies as you can. We can all agree that’s what would happen.

So, instead of glassing Tehran three times over, you take out their main missile pads, radar systems, naval units and then some. You send a strong message with a single request: drop your nuke dreams. That’s exactly what has been... politely requested since the start of the Iran war and to be fair with the Orange Man, the guy and Marco Rubio have been 100% consistent about that. Not ‘regime change’, not ‘justice for the tens of thousands protestors killed’, nope. The heart of the matter has always been nukes and nothing else. And since the bunker busters back in June 2025 didn’t finish the job, then that’s why a more ‘personal’ request had to be made.

I used to think that the IRGC were the same kind of dudes who blow themselves up in the name of religious reasons. But no, they happen to be quite smart and pragmatic about it and instead, they encourage and train and prop up other groups, like Hezbollah and the Houthis, to do the dirty bomb work for them. I guess that they like breathing more than they like the ‘death to America’ motto, so they appeal to proxies. That alone tells you that you’re dealing with people who are evil, yes, but at the same time they don’t want to end up in gibs in the name of whatever they believe.

And this, for the most part, has been obvious because of how the ceasefire has stayed put. Granted, some stray drones and missiles have broken the ceasefire here and there, but let’s not forget that upon the death of the Ayatollah, and the dubious cardboard replacement put in place, the IRGC is a group of thirty plus divisions that pretty much act on their own. So it’s not that hard to imagine that every time Iran’s president Pezeshkian is close to a deal with the Western Demons, a rogue IRGC commander decides to drive a pickup to the shore himself and launch a couple drones at Abu Dhabi. That’s not something that Tehran can actually control, and that’s by design by Tehran themselves. Trump’s words “A ceasefire in the Middle East is just shooting a bit less” (I’m paraphrasing here), are not that far from reality. The Middle East has always been extremely complex and I’m not even adding Israel and Lebanon.

So, right now I’m writing this and I’m sure that Trump must be sharing to his social media channels that either they’re really, really close to a great deal OR that he’s about to glass Tehran for good. In fact, he just said that he was about to take Kharg Island, which hosts 90% of Iran’s oil exporting capacity. I’m not going to say I have an uncanny ability to call out bluffs (I sometimes do, nevertheless), but my main question here would be, why wouldn’t you do exactly that, seize control of Kharg Island and use it as the main leverage since day one? Why wait one hundred days? Maybe there’s some kind of 7D military strategy here that someone can explain to me but I honestly can’t imagine how it could explain this in a simple way.

Which is why, understandably, people are starting to affirm that Trump is openly bluffing. I just did, in the previous paragraph. But there’s another approach, more subtle, that is Trump going step by step, ramping up the heat progressively to see if the Iran regime reacts at some point. But that wouldn’t make sense because 1) it’s 100 days of American military resources in the Arabian Gulf, 2) 100 days of the Iranian regime claiming that they’re the ‘resistance’ and they have ‘courage’ and whatever propaganda they can produce from the already stated fact that the United States has all the guns right now and 3) it’s 100 days of roller coasting headlines about the Iran war being the right thing to do or not. The markets follow suit, they have gone up and down in insane ways in the past three months and it’s not hard to imagine that whoever can predict what the headlines are going to say, is becoming quite rich about it. And since Trump is basically the main driver of headlines... well, even if you’re a Trump supporter you know there’s a huge conflict of interest in the [ generate tension -> headlines -> markets ] chain that Trump basically dominates.

Because, regardless of the outcome of the main topic of this war (Iran dropping their nuke dreams), this is going to have an actual impact in the November midterms. And Trump knows this. He’s still willing to go SLOW with Iran, and keep pushing for a deal, the same way he’s going slow with Venezuela. By the way, his latest remarks about Venezuela becoming a ‘very happy country’ and Venezuelans ‘very happy’, well, you know that message was certainly NOT for Venezuelans. That message is for Americans. Or, maybe Iranians. Because Venezuelans who are still in Venezuela have all the right to be pissed about every headline being about how ‘prosperous’ the local oil infrastructure is becoming and they’re worse than ever.

This is a good time to remind you that 1) I’m Venezuelan, 2) I don’t live in Venezuela and can’t return safely to the country and 3) I’ve always said that Trump was the miracle that the USA needed in 2024 and for my country, he’s been a net positive. I can’t expect, or no Venezuelan can expect 25+ years of corruption, drug cartels and terrorism hosting to be undone in a matter of months because I’ll tell you myself that the problem is not reds vs. blues. The problem in my country, is that reds AND most blues were on it all this time and that’s why the chavista-madurista machinery has stood for so long. Right now people want Delcy out but Delcy is the most pragmat-ish of the regime bunch so that’s the only channel Rubio can use to establish some sort of order. And that order is guaranteed by the gun pointing right at Delcy’s head, of course.

But I digress. Even so, the Venezuela example is not much different than the Iran situation. You can’t simply try to ‘glass Caracas’ and call it a day, because even if our military arm is not divided in 30 plus divisions like the IRGC, it was fractured enough to make the country absolutely ungovernable without cleaning the house first. And for Washington, a ‘clean’ Venezuela with actual elections is way more important than a regime-free Iran, because we’re in the same backyard. That’s why the Cuban regime is next.

You see, the Venezuela situation is complicated. The Iran situation is also complicated in its own way, and this is one of the reasons Trump can’t simply ‘finish the job’, because he knows that if he goes full Michael Bay on Iran, then forget about any sort of negotiation with them, and what is worse: Iran will make sure to hit back at the GCC countries, especially those who signed the Abraham Accords, not only with missiles and drones but with whatever proxies they can throw at them. And the GCC countries absolutely don’t want that. They want Iran to behave, yes; and they don’t want Iran to have nukes. In fact, nobody wants Iran to have nukes. It’s tragic what’s happening to Iranians, but in geopolitics you have to make pragmatic decisions in the best interest of your own citizens. And as long as the Iranian regime behaves (mostly) well, everyone will be OK with that.

The GCC countries understand the complexity of the situation, they want these matters to be settled, they know they NEED America’s support, and Trump knows he can’t just sweep-bomb the country. That’s why he pushes for a deal. Again and again. And that’s why he telegraphs that he’s about to hit Iran again. Maybe he will. Maybe he will actually take Kharg Island. I don’t think Iran has any capacity to retaliate against the United States or Israel, so the next best thing they can do is to inflict pain on their neighbors. I already wrote about that being the reason why they have been pummeling the UAE’s walls so hard, out of spite.

Which brings me to one thought: at this point it’s obvious that the Iranian regime REALLY, REALLY wants nukes. They’re willing to continue an asymmetric, attrition war to make sure that dream stays alive. Which in itself, should confirm what everyone thinks about them.

That’s why Trump keeps pushing for a deal. Not because he can’t land the finishing blow, but because in doing so he would precipitate a scenario way worse than the current one.

Last night an American Apache helicopter went into the water over the Strait of Hormuz. A robot boat, an actual uncrewed vessel, pulled both pilots out, which had never happened before in the entire history of people falling out of aircraft into oceans. This morning Trump blamed Iran for shooting it down and said America must respond. By this afternoon, American forces were striking Iranian military sites.

And in that same morning, before any of those strikes had landed, Trump told reporters the deal with Iran could still be signed in two or three days.

I want you to think about this for a second. A president stood in front of cameras and announced both retaliation against a country and a signing date with that same country, in the same session, with the same facial expression. I must confess that my first reaction was labeling this as a contradiction, but I have now watched enough of these to know that Trump is doing this exact thing to drive our attention away from two very critical calendars.

Israel votes at the end of October. America votes the first week of November. The two men running this war face their voters days apart, and they need the same waterway in opposite conditions. Trump needs the strait open and gasoline falling before Americans vote. Netanyahu needs the Iran question either thoroughly squashed or dramatically unresolved before Israelis vote, and since the ‘thoroughly squashed’ scenario was never really discussed by anyone, he’ll have to settle for keeping the conflict unresolved. Let me explain.

Because we’re talking about the same strait, same season, but two alarms set to opposite outcomes. But let’s take a look at their behavior instead.

Start with Trump. On Sunday night he called Netanyahu privately and asked him not to respond to Iran’s missiles. Israel struck anyway. Then Trump took his own private phone call and paraphrased his private call with Bibi. He told an Israeli journalist he had warned Netanyahu, quote, “Bibi, you better be careful, or you will be on your own very soon.” Then he repeated versions of it to reporters. Then again after a basketball game in New York, like a man determined to make sure the message reached everyone except the person it was addressed to.

There’s this thing about venting: a frustrated person does it once, in private, and feels better. Normal stuff. But Trump selected the single harshest sentence from a private call and personally hand-delivered it to three separate audiences across two days. That’s our signal. He is, very publicly, assembling a file labeled “I warned him, I asked him to stop, I gave him every chance,” and a man only assembles that file when he is preparing to do something he will need it for. You can call it an alibi, as well.

Now consider Netanyahu, whose situation is that of a man whose own war is being settled in a meeting he wasn’t even invited to. America negotiates with Iran through Pakistani and Qatari mediators, and Israel sits outside that room. A leader locked out of a negotiation has exactly one remaining way to participate in it, and that way is the battlefield. Every Israeli strike on Lebanon is meant for Hezbollah, yes; but functionally, it’s also a memo addressed to Washington, and every memo says the same thing: nothing gets signed without me.

This week his cabinet stopped pretending otherwise. His finance minister argued, on the record, for what he called the Beirut model: hit Beirut hard and keep the Lebanese front separate from the Iranian one. His national security minister announced the government cannot allow this deal to happen. And when that minister pushed for even more escalation, Netanyahu responded by suggesting the man was campaigning for the election.

So the main topic was not the Iran war, but the October elections.

The polls explain why. When Israelis were asked in April whether the Iran campaign had been a success, one in ten said yes. One in ten is a number normally reserved for questions like “do you trust airline coffee?” Two former prime ministers just merged their parties for the single purpose of removing Netanyahu, and his ultra-orthodox partners are furious over a draft law. So inventory his options. A signed deal hands his rivals the slogan “he let Iran win” right before the vote. An open war keeps him the indispensable man. The strikes are not a man losing the plot, but THE plot itself.

Two elections, days apart, pulling one strait in opposite directions. That is the machine running underneath everything you watched today.

In the following section, I am going to show you why today’s American strikes on Iran will not kill this deal, while one specific Israeli decision could. I will lay out the three ways this branches from here, what each branch does to the price of oil, and the single signal that tells you which branch we are on before the news does.

That signal has nothing to do with missiles. It sounds like this:

Read more

By the close, Marvell had gained almost a third of its value in a single day. The stock finished at 290 dollars and change. It was the biggest one-day move in the company’s history. The market value swelled by roughly fifty-six billion dollars in an afternoon. That is more than many famous companies are worth in total, conjured out of one sentence from one man.

Read more

Kuwait’s main airport was still quiet when Iran hit Terminal 1 without any warnings. A man died while thinking he was safe standing inside a civilian building meant for reunions and goodbyes, for travelers heading to family gatherings, weddings, new jobs and new experiences. A weapon launched from a country he had never touched, ended his morning and his life.

Dozens more were carried out hurt among the shattered glass, the fire climbing the walls and sirens replacing the boarding calls.

Far from the smoke, Donald Trump settled into a chair under soft studio lights and told an interviewer that the war was as good as over. Iran had already agreed to give up the bomb, he said. Calm. Certain. The voice of a man describing a deal that was all but signed.

How can both events be true at the same time?

A war can burn and wind down toward a handshake in the same breath. That impossible morning, where the weapons still fly while the deal gets drawn, is the whole Gulf right now. Once you see how both things stay true at once, everything else starts to make sense; and the best way to make sense of this mess is taking a good look at the UAE.

Back when this war opened in late February, Iran did not just hit American bases in Kuwait and Qatar and Bahrain. Iran put missiles into Al Dhafra, the big air base inside the United Arab Emirates. And Iran did not hide behind a militia in Yemen or a shadowy outfit in Iraq. The Revolutionary Guard put its own name on it. Sovereign Iran, firing on sovereign UAE, on purpose, in the open.

For years the smart-sounding take was that Iran would never dare strike the UAE directly. The story went that Iran only ever pokes the Emirates through proxies so it can shrug and say “Nah, you’re wrong, it was not us!” That story is dead. Iran crossed the line it was never supposed to cross.

But after weeks of ceasefire, Dubai is bustling again. The airport is moving people. The malls are full. The cranes are still up. The UAE kept selling its oil. The Emirates absorbed a direct hit from a sovereign neighbor and went back to business with a speed that should stop you cold, because it tells you something about how power actually works in 2026.

It tells you the thing everyone calls a red line was never a red line at all.

A red line is a rule. Rules get broken when somebody gets angry enough. Tehran proved that in February. What is actually keeping Abu Dhabi standing is not a rule and not a threat and not even the best missile shield money can buy, though the UAE has one of those too. What is keeping Abu Dhabi standing is something closer to physics.

Then came May. A drone slipped across the western border and lit a fire at the edge of the Barakah nuclear plant, the only nuclear station in the Arab world. No radiation leaked. The reactors kept running. But notice what the UAE did next: They didn’t name Iran. The UAE pointed at a drone that came in from the direction of Iraq and left the blame deliberately blurry.

So now you have two tracks running at once. On one track Iran fires openly and signs its name. On the other track a drone arrives from nowhere and nobody claims it. The Emirates take both kinds of hit and refuse to flip the table either way.

What happened to Kuwait was direct retaliation for the US Navy’s latest move in the Gulf. American forces had hit an Iranian ship and struck Revolutionary Guard sites on Qeshm Island, a small Iranian island guarding the mouth of the Strait of Hormuz, and they called it self-defense. Iran could not reach the warships that did it, so it reached for the country hosting American hardware instead. Kuwait was the mailbox. And here is the difference that matters: Kuwait cannot defend itself the way the UAE can. Worse, Iran aimed at a civilian terminal, not a base. Aiming at the place where families say goodbye is not a warning shot. It is a signal that Tehran is willing to escalate.

Because the deeper question is not why Iran would strike the UAE, or Kuwait, or Bahrain. The deeper question is why Iran cannot afford to finish the job, and why the UAE is quietly spending billions to make sure that stays true not just for oil, but for the thing that is about to matter more than oil.

Let me show you the math the Emirates are actually running, what it has to do with a channel narrower than a long drive across a city, and the one bet Abu Dhabi is placing on the next twenty years that almost nobody is naming yet:

Read more

Everything he asked for in February is in that document. The Strait of Hormuz reopens, unrestricted, no tolls, no harassment, with Iran clearing its own mines inside thirty days. Iran gives up the highly enriched uranium that started all of this. There is a sixty day window to settle the rest. By every line he drew when the first bombs fell, this is a win, and it is the most significant thing to happen since the war began.

And he is sitting on it. A couple of days to think about it, was the word that came out of the room. Meanwhile Iran is still firing drones at the very strait it just promised to open, Kuwait spent the night scrambling its air defenses against missiles coming in over the Gulf, and all of it has the shape of a regime the people around Trump quietly wonder can hold itself together long enough to sign its own name.

You have watched a version of this before, in smaller rooms. The person who holds every card goes quiet and lets the other side sweat. That quiet costs him nothing and costs the other side everything, because every hour it holds, Tehran talks itself into a worse position. The silence is the move. Every day he does not sign, the price of his signature climbs.

So the real question was never whether he signs. He will, when it is worth the most. The question is what he is buying with the wait, and the answer is not in Tehran at all.

Look at who he spent the weekend calling. Not Iran. The Arab capitals. He spent it pressing Saudi Arabia to do the thing he has wanted since his first term, to come to the table with Israel and finish what the Abraham Accords started. Iran was never the point. The cornered regime, the reopened strait, the whole region holding its breath while one man decides, all of it is the most pressure he will ever have to move the piece he actually cares about, which is in Riyadh, not Tehran. He is not going to spend that pressure on a signature one day early.

That is the behavior. A man who refuses the thing he asked for is telling you the thing was never the point.

But the diplomacy is the half you can see. The half that reaches into your life, into your rent and your gas tank and the part of your savings you try not to look at, is happening somewhere quieter, while every camera in the world points at a twenty one mile strip of water.

Because oil is doing something it is not supposed to do. There is a shooting war in the Persian Gulf. The most important oil channel on earth has been blockaded for three months. And the price of oil just fell for the second month in a row. That is not how this story is supposed to go. And what oil is quietly doing right now is telling you what the next war gets fought over, once oil stops being the prize:

Read more

Not “make good progress on Mars.” Not “land the first crew.” A million people, living up there, breathing, eating at food trucks, sharing memes and sipping lattes. And the clause is gloriously all-or-nothing, which is the part I keep turning over. Push the company to seven and a half trillion dollars and land only nine hundred thousand colonists, and he gets nothing. Zero. The contract just shrugs and waits.

Sit with that for a second, because it’s the most honest thing in the whole document. Nobody chains his entire fortune to the single most ridiculous milestone on the menu unless one of two things is true. Either he fully intends to get there. Or he has understood something about belief that the rest of us keep refusing to learn.

The economists picked the second-most-interesting reading and have been shouting it all spring. The whole thing is smoke. SpaceX wants to be valued north of two trillion dollars on sales that wouldn’t crack the list of the hundred biggest American companies, the kind of price tag that makes a Wall Street analyst’s eye start twitching. Bolted onto it is an AI business losing money faster every quarter. The major networks called it the most leveraged bet of the decade, and then, with a perfectly straight face, set aside nearly a third of the shares for ordinary people’s savings. On the math, every word of that is correct.

And that is exactly why they’re about to miss the story.

Peter Thiel, who actually sat across a table from Musk back in the PayPal days, has one hard rule he repeats like a man who learned it the expensive way. Never bet against Elon Musk, in anything. It sounds like a fan-club slogan until you notice it’s really just accounting. Every short-seller who ever called him a fraud was reading the same spreadsheet the economists are reading right now, and every one of them got carried out feet-first. Because the spreadsheet was never the thing being sold. The belief is the thing being sold.

You can watch the whole trick in a single rock. There’s an asteroid out past Mars worth, on paper, thousands of times more than the entire world economy. On paper. And every analyst who runs that delicious number also writes the very next sentence, which is that the moment you actually haul the metal home and sell it, the price collapses and the fortune evaporates in your hands. The number is real and self-cancelling at the same time. Musk doesn’t need the asteroid to pay off. He needs you to believe it might. That gap, between the impossible number and your belief in it, is the engine under everything the man has ever built.

Which brings us back to that contract, and the second prize hiding inside it, the one that should make anyone in energy or markets sit up straight. Right next to the Mars clause sits another, and Musk collects only if he ends up running data centers in orbit pushing a hundred terawatts of computing. A hundred thousand nuclear reactors’ worth of compute, humming away in space, above your head. He isn’t alone in the dream. Jeff Bezos, asked last week whether orbital data centers are real, said yes, they’re coming, just not on the giddy three-year timeline (Bezos has his own fifty-thousand-satellite version in the works, so he’s hardly a neutral witness). Blue Origin has filed to launch more than fifty thousand computing satellites. SpaceX has filed for up to a million. The logic is brutally simple. The sun never sets up there, space cools the machines for free, and there is no city council in orbit to file a complaint about the noise.

Compute is the new oil. And if these men are right, the new oilfields aren’t under anyone’s soil. They’re ABOVE it. Musk has written the address of the next century’s wealth straight into his own paycheck, and the address is orbit.

But the Mars clause isn’t the tell. The orbital compute isn’t the tell either. The real tell is the quiet decision Musk has already made about where he builds the parts of this empire that America won’t let him build at home, and once you see it, you can’t unsee where the money is about to go:

Read more

On May 26, while every camera in the world was pointed at a twenty-one-mile stretch of water in the Persian Gulf, the most important repricing of the year happened somewhere quieter. A company most people have never heard of crossed a trillion dollars in value in a single afternoon. It jumped roughly 19% in one session and added about $160 billion, more than most countries are worth, in the time it takes to watch a movie. It does not pump oil, ship it, or refine it. It makes the one thing the entire AI race cannot buy enough of.

To see why that number matters, look at what Trump spent the rest of the year doing, because he spent it fighting over the old scarce resource in three separate theaters, three different ways, and the market handed him the same answer every time.

In January, US forces seized Nicolás Maduro and took control of the country sitting on the largest proven oil reserves on earth, close to four times what the United States holds. “We built Venezuela’s oil industry,” Trump said, “and the socialist regime stole it from us.” The United States took the country with the most oil in the world. And the price of oil fell. Brent slid toward $60, because the world is not short of oil, with new barrels pouring out of Brazil, Guyana, Argentina, and Texas.

In February, the US and Israel opened a war on Iran and killed the man who had run it for thirty-six years. Iran answered by doing the thing it had always threatened and never done. It closed the Strait of Hormuz, the channel that carries almost a fifth of the world’s oil. This time the price reacted, running to $138 a barrel in April. And then, the moment a ceasefire looked plausible and the strait looked likely to reopen, it started sliding back. Brent trades near $99 this week and is on track to close the month down. The biggest supply shock in years produced a spike that is already deflating.

Then this month, the strangest move of all. Xi Jinping, the largest buyer of discounted Iranian and Venezuelan crude on the planet, got on a call with Trump, said the strait had to stay open, came out against any toll to use it, and offered to buy American oil instead. The hungriest customer for the cheap barrels turned toward the expensive, reliable ones. When your fiercest rival starts shopping for your product, the product has stopped being the prize.

Three theaters, one verdict, delivered three separate times. Seizing the largest reserve on earth could not lift the price. A war and a sealed strait lifted it for a few weeks. And the world’s hungriest buyer is now shopping for stability instead of scarcity. The oil wars are real. The oil scarcity is not.

Here is the dot that matters. The instinct driving all three moves is the oldest one in great-power politics: control the scarce resource at its source, and you control everyone downstream of it. Trump has that instinct aimed at oil. The market just told him, three separate times, that oil is not the scarce thing anymore. The scarce thing is what that trillion-dollar company makes, and its chokepoint is far tighter than any strait. Only three companies on earth can produce it. Two of them sit in a single dangerous neighborhood, and the one that crossed a trillion dollars this week is the only one of the three that answers to Washington:

Read more

Start with what almost nobody outside the energy world noticed. When Iran went looking for targets in this war, it did not only go after ships. It went after Qatar. The strike on Ras Laffan, the largest liquefied natural gas facility on the planet, took out roughly a sixth of Qatar’s export capacity in a single blow. Qatar is the third largest gas exporter on Earth. The repair estimates are not measured in weeks. They run three to five years. And the gas that was flowing out of Ras Laffan was not headed to Iran’s enemies in the abstract. It was contracted to Italy, to Belgium, to South Korea, to China. Four economies just lost a chunk of their heating, their factories, and their power generation, and there is no quick way to get it back.

Here is the part that makes this different from the oil shock everyone remembers from a few years ago. Oil is fungible and oil floats. If a barrel cannot come from one place, it can usually come from another, at a price. Natural gas does not work that way. To move gas across an ocean you have to chill it into a liquid, load it onto a specialized ship, sail it to a specialized terminal, and turn it back into gas at the other end. You cannot reroute that overnight, and you cannot conjure a new terminal in a season. When a major source goes offline, the countries that depended on it do not pay a little more. They go without.

So they are going without. The International Energy Agency has called this the single greatest energy security challenge in its history, and the response on the ground backs up the language. By the latest tracking, at least sixty countries have enacted close to two hundred emergency energy measures. Driving bans. Fuel rationing. Shortened school weeks and closed universities. Work from home orders issued not for a virus but to keep the grid from buckling. This is not happening in one unlucky region. It is happening across Asia, across parts of Europe, and across the developing economies that sit at the back of the line whenever supply gets tight.

And watch what those countries are reaching for when the gas runs short. They are reaching for coal. At least eight nations are now burning more coal or delaying the retirement of plants they had promised to close, and the list is not made up of the usual suspects alone. It includes Japan. It includes South Korea. It includes Germany and Italy. A decade of climate commitments is being quietly walked backward in a single year, because a country facing a dark winter will choose the dirty fuel it can actually get over the clean one it cannot. The lesson that every energy minister on Earth is absorbing right now is brutal and simple. Climate policy is what you do when the lights are on. Survival is what you do when they are not.

The street level picture is the part that does not make the front page. Pakistan dropped its highway speed limits to save fuel. Bangladesh asked its own citizens to switch off unnecessary lights. Laos pushed workers home. Across Southeast Asia, drivers have pulled into stations to find the diesel simply gone. None of these countries is at war. None of them fired a shot. They are just downstream of a chokepoint on the other side of the world, and the chokepoint moved.

This is the real shape of power in 2026, and it is worth saying plainly. The age of armies marching across borders to take territory is not how the strongest players are operating anymore. The move now is quieter and far more effective. You do not need to occupy a country if you can decide whether its lights turn on. Iran understood this when it drew a map instead of firing a missile. The United States understood it when it chose to choke fuel rather than land troops. The whole world is now being sorted into two groups, and the sorting is happening fast: the handful of players who can route energy around any chokepoint and keep their own lights on, and everyone else, who is learning in real time what it feels like to depend on a switch somebody else controls.

That sorting is the most important thing happening in the global economy right now, and it is quietly creating a very specific set of winners. Not the countries caught in the dark. The companies that sit on the right side of the switch.

Below the line, I lay out exactly who those are, with the names and the tickers, in five groups. The American exporters who became the world’s gas lifeline the moment Qatar went offline. The shippers earning more for the same fleet now that every cargo sails the long way around. The underwriters in London quietly collecting the war premium on every tanker that dares the strait, and the one catch in that trade most people will miss. The single cleanest way to own the UAE bypass build-out I talked about in the video, even though the pipeline itself is not for sale. And the defense names restocking the missiles the blockade is burning through. I also flag which of these is already crowded and overpriced, and the one I think the market has not woken up to yet. Here is who is positioned, and why:

Read more

That reading is wrong on every count.

Trump walked into Beijing with a public offer for ten of China’s largest tech companies:Alibaba, ByteDance, Tencent, JD-dot-com, Lenovo, Foxconn, and four others, to buy Nvidia H200 chips. The press treated it as a concession. It was nothing of the kind.

The H200 is not America’s most advanced AI chip. The B200 and the upcoming GB300 are, and Chinese companies remain completely banned from buying either one. The H200 sits one generation back. And the offer came attached to Presidential Proclamation 11002 from January 14, a 25% Section 232 tariff on advanced AI accelerators, with the H200 named specifically in the document. The actual offer to those ten Chinese firms was: permission to buy a second-tier chip, at full market price, plus an American 25% tax on top.

Trump didn’t give Xi a gift. He threw yesterday’s tomatoes on the floor and watched Xi decide whether to pick them up.

Xi couldn’t. And the reason is the part most Western analysts will not say out loud.

Beijing’s domestic chip alternative, the Huawei Ascend, has reported yield rates around 40%. Six out of every ten wafers off the production line go to the scrap pile. Every senior Chinese technologist knows it. So does every CEO of every company on that list of ten. They have been quietly begging for export-control relief for two years.

So Xi had a choice. Accept American silicon publicly, and admit, in front of the entire Chinese internet, that the Chinese Communist Party cannot build the chip it has spent a decade promising. Or refuse, and keep his data centers running on a chip that breaks more than half the time it comes off the line.

In Beijing, losing face is the actual currency the regime runs on.

Xi refused. The Trojan Horse worked exactly as designed: the gift forced the recipient to expose his own constraint by the act of declining it. And Trump (who flattered Xi throughout the trip, the warmth, the “eleven years almost twelve years,” the “incredible visit”) made the refusal cost-free for both sides. Xi declined without insulting Trump. Trump offered without spending leverage. Both walked out looking like winners.

Only one of them actually was.

The same playbook ran in a second theater the same week. Iran cannot publicly accept Trump’s no-nukes ultimatum: forty years of “Death to America” rhetoric makes a direct surrender politically impossible for Tehran. So Trump is not negotiating with the Ayatollah. He is negotiating with the Gulf states: the UAE, Saudi Arabia, Qatar, who can deliver his terms back to Tehran as a regional settlement rather than a Western victory. Same mechanic. Two capitals. One week.

What Xi actually wanted from Beijing was not chips. It was relief on the sanctions targeting Chinese companies that buy Iranian crude. Reporting from the summit confirms the two leaders discussed it. The deal couldn’t close, because every road through Iran runs into the same wall, no nukes, and that wall is held up by the Gulf, which is leaning on Tehran, which only stabilizes if Beijing helps lean harder. Xi cannot help himself unless he helps Trump first.

Beijing knows it. That is why China spent 2025 adding 500 terawatt-hours of new electricity generation (solar alone added 340) and currently has 27 nuclear reactors under construction. Alibaba just formed a joint venture with China National Nuclear Power specifically to provide dedicated electricity for AI data centers. Beijing is not building this because it cares about the environment; it is building because compute is the new oil and the indigenization clock is running.

But chip independence runs into a constraint Beijing cannot manufacture its way out of: rare earths. China controls roughly 70% of global mining and 90% of all processing. The Pentagon has spent the past year quietly building the alternative… and the compliance deadline is closer than the markets have noticed. Less than eight months. The dated catalysts, the named companies, the structural reality nobody is putting in print, and what to look for in the next 8 months is right here:

Read more

Executive summary

Trump’s Fox News line this week, telling everyone in the Taiwanese chip industry to come to America, is the political ceiling of the relocation story. The actual policy floor sits about three feet lower and has been under construction for eighteen months. The administration has built a four-piece industrial leverage architecture made up of Section 232 tariffs, bilateral capacity deals, equity-for-subsidies on Intel, and the thirteen-nation Pax Silica framework, and the whole thing is already operational and producing physical results. TSMC’s Arizona project is in volume production on 4-nanometer nodes for Nvidia and Apple, the second module enters 3-nanometer production in the second half of 2027, Amkor’s Peoria packaging campus comes online in early 2028, and by the end of this decade the Phoenix area becomes the first place outside East Asia where an advanced AI chip can be designed, fabricated, and packaged without leaving North America.

This is real, but it is also not what most people think it is. The administration is not building a fortress; it is building a network with the US at the top of a tiered hierarchy of allied producers. The rhetoric is autarky, but the actual policy is leverage, and the investment thesis underneath the rhetoric concentrates not in the chip names everyone knows but in the bottlenecks that gate the buildout: high-voltage transformers, advanced packaging capacity, specialty gases, cleanroom construction, and the announced sovereign-wealth investment consortium that may absorb Gulf capital into the next phase.

1. The leverage architecture

The legal stack now in place is more aggressive than most observers realize, and it is doing real work.

On January 14, 2026, the administration issued Presidential Proclamation 11002, imposing a 25 percent ad valorem tariff under Section 232 of the Trade Expansion Act on a narrow category of advanced semiconductor imports. The scope is narrow by design, targeting advanced AI accelerator chips meeting defined performance thresholds, hitting the Nvidia H200 and AMD MI325X by name, while exempting US data centers, startups, R&D applications, and chips imported to support the domestic supply chain buildout. This is a chisel aimed precisely at the high-margin slice of the import market, where the leverage is highest and the political cost is lowest.

The tariff is paired with a bilateral mechanism. Taiwanese firms building new fabs on US soil can export specified quantities of chips without paying Section 232 duties, and South Korea was negotiating similar terms by mid-January. The combined tariff-plus-exemption structure is the same negotiating mechanism Trump used in the first administration with Mexico and Canada, scaled up to the largest industrial dependency in the global economy, and it is working. The renegotiated US-Taiwan trade deal commits Taiwanese firms to more than $250 billion in US investment commitments, with TSMC’s own commitment now at $165 billion, expanded from the initial $65 billion announced under the Biden administration. The Arizona project has grown to six fabs, two advanced packaging facilities, and a domestic R&D center.

The equity model is the third piece. In August 2025, the administration converted Intel’s CHIPS Act grants and Secure Enclave funds into a 9.9 percent ownership stake, acquiring 433.3 million primary shares at $20.47 per share for a total investment of $8.9 billion. After Intel’s Q1 2026 earnings beat, that stake has appreciated substantially, on the order of $36 billion at recent stock prices. The administration has become an official shareholder, and it structured a completely different deal with TSMC because TSMC had leverage Intel did not. Different leverage produces different deal structures, and the two together are a working model for how to convert subsidy dollars into strategic alignment.

The fourth piece is Pax Silica, and it is the piece that almost nobody outside policy circles has fully internalized.

2. Pax Silica: why the network beats the fortress

The administration established Pax Silica at a US-led summit in Washington on December 12, 2025, as a positive-sum, multinational coalition designed to secure the AI supply chain across trusted partner nations. Founding signatories were the US, Japan, South Korea, Australia, Singapore, Israel, the UK, Greece, Qatar, and the UAE, with India joining in February 2026 and the Philippines becoming the thirteenth member in April. Under Secretary of State Jacob Helberg is the public face of the framework, which integrates critical mineral extraction, semiconductor fabrication, logistical networks, and compute infrastructure under a single diplomatic and economic umbrella.

Read more

The detail most outlets glossed past is that the interview aired the moment Trump wrapped up his summit with Xi in Beijing, and the specific arms package being held in abeyance is a fourteen billion dollar deal that had already been working its way through the channels. The framing is important. Trump is not abandoning Taiwan; he is using the timing of an unsigned arms package to apply maximum pressure on two parties at once. To Xi, the message is that the package can be slowed if Beijing delivers on the broader negotiation. To Taipei, the message is that the security umbrella is conditional on the chip relocation moving faster than the original Biden-era timeline. The chip push and the arms-sale leverage are not two stories. They are one strategy, and they were delivered in the same interview for that reason.

The major networks ran the quote, called it reckless, and moved on. The chip industry did not move on, because the chip industry has been getting ready for this for eighteen months.

Here is the part the headlines missed. Over the last year and a half, the administration has quietly built one of the most sophisticated industrial relocation programs in modern American history, and it is already operational. The tariffs are in force, the bilateral deals are signed, the equity stakes are paying off. Trump’s Fox News line is the loud version of a story that has a much quieter version running underneath, and the quiet version is more interesting than the loud one.

The leverage architecture

Start with what is already in place.

On January 14 of this year, the administration imposed a 25 percent tariff under Section 232 on a specific category of imported semiconductors. The tariff is narrow by design, targeting high-performance AI accelerator chips, with the Nvidia H200 and the AMD MI325X as the named targets, while exempting US data centers, research applications, and chips destined for the domestic supply chain buildout. The tariff is not a sledgehammer; it is a chisel aimed at one part of the import market, where the leverage is highest.

The tariff is paired with something more interesting. Taiwanese firms building new fabs on US soil can export specified quantities of chips without paying the duties, which means the tariff is the stick and the exemption is the carrot. The combined mechanism is the same one Trump used in the first administration, scaled up to the largest industrial dependency in the global economy.

It is also working. The renegotiated US-Taiwan trade deal commits Taiwanese firms to more than a quarter trillion dollars in US investment, with TSMC’s own commitment now at $165 billion. The Arizona project has expanded to six fabs, two packaging facilities, and an R&D center, with the first fab already producing chips on a 4-nanometer process for Nvidia and Apple, and the second fab entering 3-nanometer production in the second half of next year.

Then there is the equity model. The administration converted Intel’s CHIPS Act grants and Secure Enclave funds into a 9.9 percent ownership stake in the company, trading roughly nine billion dollars in subsidies for what is now a thirty-six billion dollar equity position after Intel’s last earnings beat. The administration is no longer a grant-maker; it is a shareholder, and it sat across the table from TSMC and structured a completely different deal because TSMC had leverage Intel did not.

And then there is Pax Silica, the multilateral framework the administration set up at a US-led summit in Washington on December 12 of last year. Founding signatories included the US, Japan, South Korea, Australia, Singapore, Israel, the UK, Greece, Qatar, and the UAE, with India joining in February and the Philippines becoming the thirteenth member in April. The US sits at the top of an allied production network where critical minerals come from Australia, mid-stream manufacturing happens in India and Korea, advanced fabrication centralizes in the US, and lower-margin packaging and assembly distribute to allied partners.

This is not a “bring everything home” plan. It is a “bring the highest-value piece home and put the US at the top of a hierarchy of allied producers” plan. The rhetoric is autarky; the actual policy is leverage architecture.

The three timelines

Three timelines matter when reading the chip story, and they move at very different speeds.

The political timeline is essentially complete. The deals are signed, the tariffs are in force, the equity stakes are operational, and from a story-telling perspective the “chip reshoring is happening” frame is no longer predictive but retrospective. It has already happened.

The capacity timeline is where the next four years live. TSMC’s Arizona project is on a 2027 to 2030 buildout cycle, the Amkor advanced packaging campus in Peoria comes online in early 2028 with Apple and Nvidia as anchor customers, and TSMC’s own packaging facility lands before 2029. The third Arizona fab targets 2-nanometer production by the late 2020s, while Intel’s Ohio first fab is pushed to 2030 or 2031. By the end of this decade, the Phoenix area becomes the first place outside East Asia where an advanced AI chip can be designed, fabricated, and packaged without ever leaving North America, which is a strategic milestone even if it covers only a fraction of total demand.

The structural timeline is the decades-long one. Full independence from Taiwan is not a 2030s project, because the labor pipeline takes ten years to mature, the energy infrastructure takes ten years to build, and the supplier ecosystem takes ten years to localize. What is achievable by the end of this decade is enough domestic capacity to make a Taiwan disruption survivable rather than catastrophic. Strategic insurance, not strategic autarky.

When a headline tells you the chip industry is moving to America, it is referencing the capacity timeline. When it tells you full self-sufficiency is impossible, it is referencing the structural timeline. Both are true at once.

The constraints that do not bend

The physical constraints on the capacity timeline are the part of the story that does not move at the speed of statements.

Energy is the hardest one. Advanced fabs need 100 to 200 megawatts each, around the clock, with no flexibility, because a power dip ruins a wafer. The US grid has more than 2,100 gigawatts of generation projects sitting in interconnection queues, which is more than the entire installed capacity of the current US grid, all waiting in line to plug in. Transformer lead times have stretched from 24 months in 2019 to 5 years in 2026, and half of the AI data center projects announced for this year are projected to slip to 2028 or never get built. The chip fab buildout and the AI data center buildout are on a collision course, both demanding the same firm dispatchable power, in the same parts of the country, on the same timeline.

Labor is the next hardest constraint. The US needs roughly 67,000 more semiconductor workers by 2030 than are projected to exist, and more than half of the master’s engineering graduates at US universities are foreign citizens, three-quarters of whom leave the country after graduation. TSMC’s first three-year rotation of Taiwanese engineers in Arizona is approaching the end of its term, and the visa policy the administration is running is in active tension with the staffing model TSMC needs. The resolution is probably a chip-industry-specific visa carve-out, but it has not happened yet.

Helium is the quiet one. The Strait of Hormuz disruption in March took roughly a third of global helium output offline, and EUV lithography (which is what makes the most advanced chips possible) depends on helium for cooling and leak detection. The semiconductor industry uses a quarter of global helium production, which means a sixty to ninety day Strait closure starts to bite hard. The Australia critical minerals framework was the first move in what will probably become a similar framework for noble gases.

These constraints do not invalidate the relocation story; they define its pace. The policy machinery has moved as fast as paper and signatures allow, but the transformers, the engineers, and the helium tanks move at the speed of physics.

What the headlines are not connecting

Most retail readers are still anchored on the obvious plays: buy TSMC, buy Nvidia, buy Intel. Those are the names everyone knows, and they are also the names that already reflect the relocation thesis in their price.

The asymmetric plays are downstream of the bottlenecks, not in the names everyone is staring at. The grid bottleneck has names, the packaging buildout has names, the materials chain has names, and the equipment suppliers get paid no matter who wins the fab race. And then there is the announced four trillion dollar Pax Silica investment consortium, with named Gulf sovereign-wealth participation that almost nobody is talking about yet.

The names sit underneath the rhetoric, not on top of it. The six worth watching are:

Read more

Then Trump sat down with Sean Hannity, and what came out went well beyond the trip readouts. Xi went past the oil-purchase line. He explicitly offered to help Trump open the Strait of Hormuz and broker an Iran deal.

Trump’s quote, on tape: “Xi would like to see a deal. He said, ‘If I can be of ANY help at all, I would like to be of help.’” And then the line that actually matters, where Trump added: “anybody that buys that much oil has obviously got some kind of a relationship with them.”

That is Trump telling the audience, in plain language, that Xi has just offered to lean on the Iranian regime he has spent twenty years propping up.

The video walks through the surface of the trip. This briefing is for the layer underneath. What Xi is actually offering, why he is offering it now, what Iran has just lost, and where the receipts will show up first.

The first question worth sitting with, before anything else, is this:

Read more

That was the video.

What did not fit in twelve minutes is a second story sitting underneath all of it. On May 9, four days before Trump landed in Beijing, Iran’s state media published a regulatory proposal most observers passed over entirely.

It was a framework for seven undersea fiber-optic cables.

The Cables

The seven undersea fiber-optic cables passing through the Strait of Hormuz carry a meaningful share of global internet traffic between Asia, the Persian Gulf, Europe, and East Africa. Most international communications between South Asian data centers and European endpoints route through this corridor. The cables sit on the seabed in waters Iran controls or contests.

On May 9, state media linked to the Iranian Revolutionary Guard published a detailed regulatory proposal. Iran wants permits for any traffic transiting these cables. Iran wants fees collected by Iranian companies. Iran wants Iranian regulators with technical access to the routing infrastructure.

This is the digital version of the Hormuz tanker tolls Tehran has floated for years on the oil side. Same chokepoint, different commodity. The commodity this time is bandwidth.

Now, what does this have to do with the Chip story?

Compute requires three things. It requires chips. It requires energy. It requires data movement. The video covered the chip side and the energy side. What it did not cover is that compute means nothing if the data cannot move.

Training a frontier AI model is not a single-location operation. Training data flows in from dozens of sources. Model checkpoints flow out to inference clusters. Distributed training across data centers in different countries is increasingly common as power and cooling constraints fragment the industry. All of it requires high-bandwidth links between continents.

If Iran imposes a regulatory framework on the Hormuz cables, every Asian AI company sending data to a European inference cluster pays an Iranian toll. The alternatives are longer, slower, or more politically exposed. The Suez routes are increasingly contested. The Pacific routes have their own bottlenecks at the Luzon Strait.

Trump controls one chokepoint at the hardware layer through the Foster Bill. Iran is positioning to control the other at the data layer. Two chokepoints, one resource, and a clock running on both:

Read more

That argument is correct as far as it goes. The summit ends Friday. The reset already in motion outlasts it by a decade. What follows is what comes next in three specific places, and why one of them decides who powers artificial intelligence by 2030.

Let me start with the smallest of the three pieces, which is also the loudest this week.

“Cuba is asking for help.” That phrase has a specific meaning in regime-collapse language, and the meaning is not “we want diplomatic dialogue.”

A regime asks for help when the lights go off in the hospitals. When the dialysis machines stop. When the rice stops moving from the port to the bodega. When the soldiers stop getting paid in food. The Cuban regime survived the Soviet collapse in 1991 because Venezuela showed up two years later with discounted oil. Cuba survived the 2014 oil price crash because Maduro doubled the discount instead of cutting it. Cuba survived the pandemic because Venezuelan shipments held.

In every one of those crises, the lifeline was Caracas, paid for in oil.

Caracas no longer sends oil at a discount. Caracas no longer sends oil. The man who ran the discount system is in a Brooklyn jail in a beige jumpsuit. The Venezuelan oil ministry that used to sign off on the Cuban shipments works under a new government now, and that new government has a different list of priorities. Cuba is not on it.

So when Trump posts that Cuba is asking for help, he is signaling that the regime crossed the threshold where the only options left are to negotiate with Washington or watch the lights go off permanently. Trump did not choose this outcome. The thermodynamics of the regime chose it. The Castros’ system was built on cheap subsidized hydrocarbons from somewhere else. There is no longer a somewhere else.

The timeline most analysts are not pricing in: Cuba is closer to the 1989 East Germany scenario than the 2019 Venezuela one. East Germany did not negotiate a soft landing. East Germany ran out of money, ran out of fuel, ran out of legitimacy, and the regime ended in nine months. Cuba’s running clock looks closer to that than to a decade of slow agony.

What “after the regime” looks like in Cuba matters for one specific reason that has nothing to do with Cuba. A non-hostile Cuba removes the Russian submarine problem from the American security perimeter. A non-hostile Cuba opens the Caribbean to American naval projection in a way that has not been possible since 1959. A non-hostile Cuba lets the United States move energy and trade through the Yucatán Channel and the Straits of Florida without watching the radar every twelve hours.

That is the piece most analysts will write about this week.

It is also the smallest of the three pieces.

Cuba is the loud story. The two quieter ones decide who runs the next decade, and one of them is happening in a country most analysts have given up on for ten years:

Read more

Tonight’s video makes a simple argument: Trump is not flying to Beijing as the supplicant. Xi is the one who needs the meeting. The inflation print, the thirty-percent collapse in Chinese oil imports, and the choice of which executives Trump brought on the plane all point in the same direction.

That is the surface read. It is correct as far as it goes.

But the surface read leaves three pieces on the table that the video did not have room to walk through. Each one is the kind of dot most analysts will not connect this week, because connecting them requires holding four storylines in your head at the same time and asking what they look like from the inside of the room where the decisions are getting made.

The first piece dropped a few hours ago and the timing is not an accident…

What the Wall Street Journal Just Confirmed

The Wall Street Journal reported tonight that the UAE secretly carried out military strikes inside Iran. Specifically, the April 8 attack on the oil refinery on Lavan Island, a fifty-five-thousand-barrel-per-day facility that Iranian state media at the time described only as an “enemy attack” with no attribution.

The Journal links the strike to the UAE. The aircraft are believed to have been Mirage 2000-9s. The Emirates have not publicly acknowledged the operation. The strike took place hours after a US-brokered ceasefire was signed. Iran responded immediately with missile and drone attacks on UAE targets.

For weeks the running framing of the Iran war has been that the United States is conducting the kinetic operations and the Gulf states are providing basing, intelligence, and political cover. The Wall Street Journal just rewrote that frame.

What the Journal actually reported is that the UAE has been running a parallel campaign against Iranian oil infrastructure, with Western-supplied aircraft, while the United States manages the diplomatic and naval pressure. Two campaigns. Two angles of attack. One coordinated objective.

This is the part that matters for the Beijing summit:

When Trump sits down with Xi on Wednesday, the leverage in the room is not just the seventy tankers blockaded by CENTCOM. It is the fact that even if Trump backed off tomorrow, even if he agreed to ease the naval pressure as part of a deal, the UAE would keep going. And Israel would keep going. And Saudi Arabia, which has been notably quiet through this entire war, would not lift a finger to stop them.

That is a structural feature of the chokehold that Xi cannot negotiate away. He can ask Trump to call off the US Navy. He cannot ask Trump to call off Abu Dhabi.

Which means the offer Trump is bringing to the table on Wednesday is not just I will turn the oil back on. The offer is I will turn the oil back on, and I will also persuade the Gulf states to stop hitting Iranian refineries from the air. Two switches. Both in Trump’s hand. Neither in Xi’s.

And there is something else the Beijing summit is going to have to decide that almost nobody outside a handful of analysts is paying attention to right now.

It involves a Chinese AI company that just raised forty-five billion dollars from China’s state semiconductor fund, and it is not the story the financial press is telling about it.

But to understand what DeepSeek is actually doing in the background, why the seven-trillion-dollar rare earth threat is a paper tiger, and what Wednesday's meeting will actually decide, keep reading…

Read more

The United Arab Emirates has assembled four heavy-hitting, foreign-policy relationships: Washington, New Delhi, Tel Aviv and Beijing into a single working system that the artificial intelligence economy now physically depends on.

Every other country chasing the same role has, at most, two of those relationships.

Most have one.

Take Saudi Arabia, which has the money. Or Singapore, with the neutrality. Ireland has the legal framework. I could keep going for a while.

But the UAE has all of it at once, and it’s been quietly delivering against the plan while everyone else has been holding press conferences. The best part is that you're about to step inside what the Western press has not even noticed yet.

And by the time it does, the concrete will already be poured.

A short tour of why this even matters

Before going further, here is the thing about modern AI that most casual observers do not yet understand: it is a heavy industry now. The cute years are over. “AI is a bubble” used to be a surefire click in the first half of the 2020s.

Not anymore.

The main problem is that training a frontier AI model, like ChatGPT, or Grok, or Claude, requires a building roughly the size of a shopping mall, packed wall to wall with specialized computer chips, cooled by enough water and electricity to power a small city.

These buildings are called data centers. The most advanced ones are called AI campuses. They cost tens of billions of dollars each, take years to build, and consume gigawatts of electricity, where one gigawatt is roughly the output of a full nuclear reactor.

This means the AI race is no longer really about who has the smartest software engineers. The software is converging and turning into a commodity, while the infrastructure they run on is turning into the real moat. (I suggest you also read my briefing Why Compute Is the New Oil)

The infrastructure dilemma includes 1) chips, which are made almost exclusively by Nvidia in the United States and TSMC in Taiwan, and 2) electricity in industrial quantities, which poses its own problem.

But the actual strategy is figuring out the political relationships to keep both flowing, in a world where the United States is choking China’s chip access, China is hoarding the metals that go into the chips, and almost every country in between is being asked to pick a side.

The country that puts those three things together first, wins the decade in front of us.

And this is where the UAE comes in…

What the construction site actually looks like

In May 2025, on the margins of Donald Trump’s tour through the Gulf, the United States and the UAE signed a framework that authorized the export of up to five hundred thousand of Nvidia’s most advanced chips to the UAE every year through 2027. To put that number in scale: that is more high-end AI chips than most nations on earth will ever see in total, going to a country of around eleven million people, every twelve months, for three years.

Twenty percent of the chips were earmarked for an Abu Dhabi-based AI champion called G42. The remaining eighty percent went to American technology giants: Microsoft, Oracle, Nvidia and OpenAI, to operate inside the UAE.

Alongside the chip framework, the same trip announced Stargate UAE: a five-gigawatt artificial intelligence campus in Abu Dhabi, built by G42 in partnership with OpenAI, Microsoft, Oracle, Cisco, Nvidia, and SoftBank. Five gigawatts is roughly the electricity demand of the entire city of Boston. The campus covers ten square miles, which is about nine times the surface area of Monaco. It is the largest AI facility being built anywhere outside the continental United States.

By the end of last year, the chairman of Mubadala, the UAE sovereign fund that helped seed G42, was telling an audience at the Milken Summit that the project had gone “from five kilometers per hour to two hundred and fifty.” The CEO of G42 told reporters in January that there were already more than seven thousand workers on site, with a hundred cranes in the air and steelwork weighing one and a half times the Eiffel Tower. The first two hundred megawatts come online this autumn. The full first gigawatt finishes within three years.

Meanwhile, the European Union is still figuring out how long its list of AI regulations should be.

The American relationship

A short history of G42 helps explain why the American relationship matters so much. As recently as 2023, Congress was sending letters about G42’s links to Chinese companies: Huawei in telecom hardware, ByteDance in social media. By February 2024, every one of those links had been removed. The ByteDance investment was sold. The Huawei equipment was physically pulled out of the UAE’s data centers and replaced with American hardware. The CEO of Microsoft’s regional operations took a board seat. Microsoft itself put one and a half billion dollars on the table.

What replaced the old setup is something the American government had been hoping to design for years and had never managed to build elsewhere: a verifiable system for moving frontier AI technology to a foreign partner with cryptographic tracking on every chip, end-user verification on every workload, and a compliance regime that satisfies American security agencies. The framework now has its own name: the Regulated Technology Environment, and is being held up in House Foreign Affairs Committee testimony as the model the rest of the world should adopt.

The capital side moved at the same speed. The Mubadala-backed AI fund called MGX joined a partnership with BlackRock, Microsoft, Nvidia, and Elon Musk’s xAI to build AI infrastructure across the United States and allied countries. In October 2025 that partnership took private a company called Aligned Data Centers for forty billion dollars. It was the largest data center deal ever made. Inside that single year, Mubadala alone deployed roughly five billion dollars into AI investments, more than any other government-controlled fund anywhere on earth.

The Indian relationship

In a Mumbai banquet hall in September 2024, G42 announced that it was going to build two gigawatts of data center capacity inside India. To put that into context, two gigawatts was roughly twice India’s entire existing AI compute capacity at the time. At the same event, G42’s research arm released a thirteen-billion-parameter Hindi-native AI model called NANDA. Prime Minister Modi had personally signed the framework agreement during his earlier UAE visit.

The relationship runs much deeper than any single deal. There are roughly four million Indian nationals living in the UAE, which is about thirty-eight percent of the population. They run the country’s tech sector at the staff level, the engineering level, the small-business level. The two countries share a free-trade agreement that has been in effect since 2022 and covers more than a hundred service categories, including digital services and computing.

There is also a corridor under construction called the India-Middle East-Europe Economic Corridor. It froze during the Gaza war and was restarted by Modi’s speech to the Israeli Knesset in February. Today, with the Strait of Hormuz periodically blocked by the ongoing war with Iran, that corridor is being actively used as a shipping bypass. It is the only land-and-sea route from India to Europe that does not run through hostile water.

What India does not have is a domestic chip manufacturing base or an ally that can route Nvidia’s frontier silicon to it through legal channels at scale. What the UAE has is exactly that.

Every one of the next billion compute users coming online in South Asia will, in some practical sense, be sitting downstream of an Abu Dhabi data center decision.

The Israeli relationship

The Abraham Accords — the 2020 normalization agreements between Israel and several Arab states — were widely expected to fall apart in October 2023.

Spoiler alert: They didn’t.

UAE-Israel trade reached three and a half billion dollars in 2025. Cross-border private technology investment rose four hundred and thirty percent year over year. The Israeli defense company Elbit Systems signed a two-point-three-billion-dollar air-defense contract with the UAE late last year, with technology transfer phased in over eight years. The Mohamed bin Zayed University of Artificial Intelligence runs a joint research program with Israel’s Weizmann Institute. Early-warning radar data flows in both directions on a hardened technical bridge.

This held through a war. It held through the Hamas attack, the Gaza campaign, the regional escalation, and the current Iran war. The relationship survived because senior people on both sides treated it as infrastructure rather than ceremony: quietly, technically, and without the public optics that would have forced one side or the other to break.

The Chinese relationship

This is the piece most American analysts get wrong, because they read the headlines about chip controls and assume that the UAE picked a side. The UAE did not pick a side. The UAE built a wall.

G42, the AI vehicle, is firewalled from the Chinese technology stack at every layer the Americans care about. Everything else in the relationship is intact and working. UAE non-oil trade with China crossed one hundred billion dollars in 2024. ADNOC, the UAE’s state oil company, ships roughly seven hundred thousand barrels of crude per day to China, which amounts to about six percent of all Chinese seaborne crude imports. The Chinese state oil companies retain stakes in UAE oil concessions. The terminal at Khalifa Port is operated by COSCO, the Chinese state shipping line. The chairman of Mubadala — the same Mubadala that runs MGX, the same MGX that just bought a forty-billion-dollar American company, sits on the advisory board of Tsinghua University’s School of Economics. He also serves as the UAE’s special envoy to China. In April, the next ruler of Abu Dhabi flew to Beijing personally.

This is what an actual geopolitical hedge looks like when adults are running it. The Chinese understand exactly what is happening at G42. The Chinese have not retaliated, because the rest of the relationship is worth keeping. When Iran threatened to escalate this spring, China’s foreign minister publicly urged Tehran to “pay attention to its neighbors’ reasonable concerns”. A quiet message that Beijing values Gulf energy stability over solidarity with Iran. The UAE bought that posture, in part, by keeping the door open everywhere else.

This is the trick. This is the trick that nobody else can do, and many, many have missed.

Why nobody else can copy this

Saudi Arabia is the country most often mentioned as a competitor and is in many ways an admirable one. Its sovereign fund is large. Its AI vehicle, called HUMAIN, is now on the field. It has launched serious technology partnerships with serious American companies. The reason it cannot replicate the UAE stack is mostly about timing. The UAE began this work fifteen years ago, when nobody was watching. Saudi began three years ago, when everybody is. That gap matters because the diplomatic relationships that hold the alliance stack together cannot be acquired with capital. They have to be built across decades, through small consistent gestures, through patient envoys, through universities that send their best graduates to study in each other’s countries. The UAE did that. Now Saudi is doing it. The work is real, and the gap is real, and both things can be true.

Singapore is a hub, but it is small and has no comparable energy base. Ireland has the legal scaffolding for data residency, but it froze its data center permits in 2022 because the national grid cannot support the load. Nobody else is even in the conversation.

The two pillars under the spine

The four-arrow alliance is the spine of the strategy. It rests on two pillars that the Western press routinely mistakes for the strategy itself.

The first pillar is capital. The Abu Dhabi sovereign cluster — the funds called ADIA, Mubadala, ADQ, IHC, and MGX — manages somewhere around two and a half trillion dollars. That is roughly the gross domestic product of France. These funds are coordinated through a council called the Artificial Intelligence and Advanced Technology Council, with weekly decision authority sitting inside a room of perhaps five men. Capital with that kind of governance and that kind of horizon does things American private capital structurally cannot. It funded a sixteen-year nuclear power program, the Barakah plant, that is now generating a quarter of the UAE’s electricity carbon-free. It funded a fifteen-year semiconductor company, GlobalFoundries, that is now critical to the American chip supply. It funded a twenty-year renewable energy company, Masdar, that is now the biggest single foreign investor in American clean power.

The second pillar is energy, and the man speaking publicly about it most consistently is Dr. Sultan Al Jaber, who is the UAE’s Minister of Industry, the chief executive of ADNOC, and the chairman of XRG, ADNOC’s international energy investment arm. At the Atlantic Council Global Energy Forum in June 2025, he said the relevant sentence: “The race for AI is not just about code. It’s about gigawatts.” The UAE has Barakah, delivering five point six gigawatts of carbon-free baseload. It has Masdar’s domestic and international solar deployments. It has natural gas at the wellhead. It has a permit-to-power timeline of under two years for new compute capacity, against three years and counting in the United States and a permit freeze in Dublin. While other countries debate, the UAE pours concrete.

The closing signal

In December 2025, the UAE and Israel founded a framework called Pax Silica — a trusted-supply-chain agreement for advanced AI chips, with cryptographic verification and reciprocal commitments. The United States joined as a senior partner. In February of this year, India joined as the fourth founding signatory.

Stop and look at that list. The United States. The UAE. Israel. India.

Those are the four arrows of the alliance stack, no longer four separate bilateral relationships, now joined into a single supply-chain architecture for the most strategically important industry of the century. The UAE is the central routing node. This is what fifteen years of patient diplomacy looks like at the moment of execution.

The Western press is still asking whether the UAE belongs in the AI race at all; the question they should be asking is whether anyone running the AI race can afford not to route through them.

The chips, the gigawatts, and the alliances are the three things. The chips are mostly settled. The alliances are mostly built. Which leaves one variable still moving fast enough to forecast against, and that is where the next two years are going to be won or lost:

Read more

Trump launched Project Freedom overnight, and starting Monday morning Middle East time, US Navy guided-missile destroyers begin escorting foreign-flagged commercial ships through the strait under direct presidential authorization.

Within eight hours of the announcement, the Iranian Revolutionary Guard fired at two ships on opposite shores: a bulk carrier off Sirik on the Iranian side, and an empty Emirati tanker called Barakah near Fujairah. The UAE Defense Ministry confirmed it intercepted four Iranian cruise missiles over Dubai and Sharjah, the first missile alert on Emirati soil since the April ceasefire. Iranian state media claimed the IRGC stopped a US warship and hit a US Navy patrol boat with two missiles.

CENTCOM denied both within hours.

The mainstream media is filing escalation stories, but the actual frame, as usual, is the opposite. The IRGC fired drones at an empty tanker and missiles into the Gulf because the regime has nothing else left, and the American escort operation just told every neutral country on earth that Washington is the guarantor of global shipping in this waterway, not Tehran.

The numbers behind the kinetic noise are the story. CENTCOM’s commander said today that the blockade has now stranded forty-one tankers carrying roughly sixty-nine million barrels of unsellable Iranian crude, somewhere in the range of six billion dollars of oil that nobody can buy and nobody can deliver. The Apache flyover of the strait the day before Project Freedom was totally a walk-through.

But the harder story landed away from the cameras.

Saudi Arabia today restored its East-West pipeline and, by regional energy desk reporting, is now pumping something close to its full export book across the Arabian peninsula to Red Sea ports, bypassing the Strait of Hormuz entirely. Combine that with the United Arab Emirates exiting OPEC last week, the UAE-South Korea Economic Partnership Agreement entering into force today (eliminating tariffs on most goods, the first such deal between South Korea and any country in the Middle East), and Trump rejecting Iran’s fourteen-point peace plan on camera as “not acceptable.” The Saudi-Emirati-American-Korean corridor that the war was supposed to put at risk is being formalized in writing, in public, in the same week the Iranian regime is firing missiles at empty tankers.

The chokepoint Iran spent forty years building its entire foreign policy around is being deleted by infrastructure, by trade agreements, and by escort operations, simultaneously. That is the Hormuz story. It is the loud one, and it is the easier of the three to read.

The other two stories were quieter. One of them is sitting inside the Pentagon, and it explains why this corridor matters more than the war that built it:

Read more

A Chinese lab released a model in 2025 that mimicked the architecture and behavior of one of Anthropic’s frontier systems closely enough that researchers noticed within days.

Open-weight models from labs in France, China, and the United States now sit within months of the proprietary frontier on most benchmarks. The agent layers that sit on top of these models are converging too. The same patterns appear in every research paper, the same training tricks, the same scaffolds, the same evaluation protocols.

This is the shape of the technology itself, so no AI company has an actual ‘moat’, if we’re just talking about the models.

Software, given enough time, gets copied. The first generation of any breakthrough is rare. The fifth generation is everywhere. Anyone tracking the history of databases, web browsers, search engines, mobile operating systems, or cloud infrastructure has seen this pattern play out. The novel becomes standard. The standard becomes free. The free becomes a commodity.

The artificial intelligence race is collapsing into the same pattern at a speed nobody expected.

That sounds like bad news for whoever is currently leading the model layer, but in fact, it is just a relocation of the real prize.

If models stop being the differentiator, something else has to be. And the something else is already visible to anyone willing to look one layer down.

The race is for compute. The compute is built on chips. The chips depend on rare metals. All of it runs on electricity.

Whoever controls the physical stack controls the next twenty years of artificial intelligence, and through it, the next twenty years of national power.

Why models are becoming commodities

A frontier model in 2026 looks roughly the same as a frontier model from 2027 will look, just somewhat better. The architectures are converging. The training recipes are leaking. The talent is mobile. A senior researcher at a top lab can move to a competitor and bring most of what they know with them, regardless of non-disclosure agreements.

Ideas do not stay in safes.

What does not move easily is what built the model in the first place: tens of thousands of advanced chips, gigawatts of electricity, billions of dollars in capital, and the supply chains that fed all three.

Picture two countries trying to build the same skyscraper. Both have the blueprints. Both have the architects. One has steel mills, copper smelters, cement factories, and a working power grid. The other has none of those things and has to import them under sanctions.

The blueprints are equal, but the ability to build is not.

The blueprints are AI models. Having them is great, but the actual race is much more physical.

This is why the most consequential decisions of the next decade are no longer being made in research papers. They are being made in trade negotiations, mining rights, electricity contracts, and chip export licenses.

Compute is the new oil

For most of the twentieth century, oil determined which countries had freedom of action. A nation that controlled enough oil could feed its military, run its economy, and project power abroad. A nation that had to import oil was always one embargo away from collapse. The Cold War, the Gulf wars, the relationship between Saudi Arabia and Washington, the rise of Russia as a petrostate, the structural fragility of post-1979 Iran. Oil was not the only factor in any of those stories, but it was always present, and often decisive.

Compute is becoming the same kind of strategic resource for the twenty-first century.

A nation with abundant compute can train better models, run more agents, simulate more scenarios, predict more events, and automate more decisions than a nation without it. The advantage compounds. A government that uses artificial intelligence to allocate capital, target weapons, gather intelligence, and run logistics better than its rival will outpace that rival by orders of magnitude over a decade, not percentages over a year.

This is why the largest AI infrastructure projects on Earth are being announced at a cadence that would have looked absurd two years ago. A single campus under construction in the Persian Gulf is rated for several gigawatts of power.

The combined American buildout, often referred to under the umbrella of the Stargate program, is targeting hundreds of billions of dollars in compute infrastructure over a five year window. Chinese state planners are matching the announcements with their own.

Compute is now treated by every major government as critical national infrastructure, in the same legal and regulatory category as power generation, telecommunications, and weapons production. The category change is recent. The implications are still being worked out.

The chip layer

A modern artificial intelligence chip is the most complex object humans manufacture at scale. The advanced graphics processors that train and run frontier models contain transistors at sizes measured in single-digit nanometers, smaller than the wavelength of visible light. A single such chip requires hundreds of manufacturing steps using equipment built by perhaps a dozen specialized firms in the world.

The most consequential of those firms sits in the Netherlands and makes the lithography machines that print the smallest features. The most consequential foundry sits in Taiwan and runs the bulk of the world’s advanced chip production. The most consequential designer of artificial intelligence chips sits in California and depends entirely on the foundry in Taiwan to actually manufacture its products.

Three countries hold the keys to the entire frontier. Pull any one of them out of the chain and the global frontier of artificial intelligence stops moving.

This is the geometry that makes Taiwan the single most important geopolitical pressure point on Earth in 2026.

It is why the United States has spent the past five years subsidizing the construction of foundry capacity in Arizona and pressuring its allies to host parallel facilities. It is why the Chinese leadership treats reunification with Taiwan not as a sentimental ambition but as a strategic necessity.

The country that controls the foundry controls the chip.

The country that controls the chip controls the model.

The country that controls the model… controls the rest.

The Americans currently hold the design layer and the lithography layer through allies. The Chinese are racing to build a parallel domestic supply chain that does not depend on either. The race is years behind on the most advanced nodes and closing fast on the older ones.

Whoever closes the gap first, changes the entire balance.

The rare earth layer

Underneath the chips is a layer most readers outside the industry have never heard of. It is the layer that makes the chips, the magnets, the lasers, the optical fibers, the radar systems and the electric motors physically possible.

The processed feedstock for that layer is dominated by China to a degree that has no parallel in any other strategic resource.

Roughly four out of every five tons of refined rare earth elements consumed globally pass through Chinese processing facilities. For some specific elements, the Chinese share approaches the totality of world supply. Neodymium and dysprosium, which together make the high-strength permanent magnets inside electric vehicles, wind turbines, and precision-guided weapons, are largely processed in two provinces in southern China. Gallium and germanium, which are essential to advanced chip packaging and certain laser systems, were placed under Chinese export licensing in 2023. The licensing regime has expanded since.

This is not an accident. The Chinese government identified rare earth processing as a strategic priority in the late 1980s and spent thirty years building the infrastructure, while Western governments shut down their own processing facilities under environmental pressure. The resulting asymmetry is the kind of vulnerability that gets noticed only when it is too late to fix in a single political cycle.

The Americans are building processing capacity in Texas. Australia is expanding mining and partial processing. The European Union is funding strategic stockpiles. The Saudis and Emiratis are bidding for ownership stakes in African mineral plays. None of these projects will produce significant volume before 2028. Most will not produce significant volume before 2030.

In the meantime, the country that processes the metals can also withhold them. Every advanced chip, every electric vehicle and every guided missile sits downstream of that decision.

The energy layer

The final layer is the one that makes everything else possible: electricity.

A modern data center campus capable of training frontier models draws somewhere between several hundred megawatts and several gigawatts of continuous power. A single gigawatt is roughly the output of a large nuclear reactor, or the electrical demand of a city of a million people. The largest campuses currently announced will draw the equivalent of medium-sized nations.

The electricity has to come from somewhere, and it has to be cheap, abundant, and reliable. Intermittent sources like solar and wind, on their own, do not work for training runs that have to operate continuously for months. Batteries help but cannot bridge the gap at the required scale. The realistic options are nuclear, natural gas, hydroelectric, and in some places geothermal.

This is why the most aggressive infrastructure buildouts are happening in three categories of countries. The first is the United States, which is reopening retired nuclear plants, fast-tracking new natural gas turbines, and approving small modular reactor designs at a pace not seen in decades. The second is the Persian Gulf, where natural gas is abundant, sovereign capital is patient, and regulatory approvals can happen in months rather than years. The third is China, which is building nuclear capacity faster than the rest of the world combined and treats grid reliability as a national security matter.

The countries that cannot bring power online quickly will rent compute from the countries that can. The rent is not just financial. It is informational, strategic, and political.

What an AI-enhanced government actually looks like

The phrase “artificial intelligence” still suggests, to most people, a chatbot. The actual capabilities being deployed at scale by serious governments are something else.

Military targeting systems are using machine learning to identify, track, and prioritize objectives in real time across satellite, radio, and signals intelligence feeds. The window between detection and engagement is collapsing from minutes to seconds.

Economic forecasting and sanctions modeling departments inside major treasuries are running large language models against entire archives of trade data, shipping records, and financial flows to find evasion patterns that human analysts would never catch.

Foreign ministries are running scenario simulations against summit positions, treaty texts, and rival statements to anticipate counterparty moves before they happen.

Domestic security services are running facial recognition, gait recognition, and language analysis against population-scale data to track individuals of interest.

Information operations groups are generating synthetic content, translating it into dozens of languages, and distributing it through automated networks at a volume that drowns out organic communication on contested topics.

These capabilities scale with compute. A government with twice the compute can run twice the simulations, train twice the models, surveil twice the population, and generate twice the content. The advantages compound across decades.

The countries that fall behind on compute will not lose dramatically in any single moment. They will lose slowly, in every negotiation, every market move, every military exchange, every information cycle. The accumulated disadvantage is what matters. Once it sets in, it is almost impossible to reverse inside a single political generation, because the compute gap that produced it has only widened in the time it would take to respond.

The map

The current race breaks roughly along the following lines.

The United States holds the deepest model labs, the most advanced chip designs, the largest pool of patient capital, and a network of allies who control the most critical foundry capacity in Taiwan, the most critical lithography supplier in the Netherlands, and significant portions of the rare earth mining base in Australia and Canada. It does not have processing capacity for most rare earths. It does not have abundant cheap electricity in every region where compute is being built. It depends on Taiwan in ways that no responsible national security planner would accept if there were any alternative.

China holds the processing capacity for nearly every critical metal, an unmatched manufacturing base for the components that surround the chips, the sovereign ability to coordinate buildout across decades, and the fastest electricity generation expansion of any major economy. It does not yet have the most advanced chip designs or the lithography to produce them domestically. Its model labs have closed most of the gap on the open-weight frontier and are still chasing on the proprietary frontier, but the gap is measured in months now, not years.

The Persian Gulf states, especially the United Arab Emirates and Saudi Arabia, hold a unique position. They have abundant cheap energy, sovereign capital, neutral political alignment between the two giants, and a willingness to host infrastructure both sides need. Their bet is that hosting the physical layer makes them indispensable to whoever wins the software layer. So far the bet is paying off.

Europe has the lithography, partial chip design talent, regulatory weight, and almost none of the energy or processing capacity. Its current trajectory points toward becoming a customer rather than a player.

Russia has energy, raw materials, and a small but real research base. It has been excluded from the most advanced chip supply chains and is unlikely to close that gap inside the relevant window.

Every other country either rents compute from one of the above or accepts that it will not be a serious actor in this race.

The conclusion

The race the public sees is the race for better models. Each new release is announced with fanfare, scored against benchmarks, and celebrated or criticized depending on the audience. That race is real, and it matters, but it is also closing on itself. The models are becoming commodities. The performance gaps are narrowing. The architectures are converging. Five years from now the leaderboard at the top of the open-weight rankings will be crowded, and the difference between the best proprietary model and the best open one will be small enough to argue about.

The race that decides the next twenty years is happening underneath. It is being fought over chips, the metals that build the chips, the machines that print the chips, the foundries that manufacture them, and the gigawatts of continuous electricity that the chips require to do anything useful.

The countries that solve the physical stack will set the pace of artificial intelligence development, deploy the most capable systems across their militaries and economies, and exert the strongest influence on every other country that needs to rent what they have built. The countries that fail to solve the physical stack will become customers, dependents, and eventually clients of the ones that did.

This is not a software race. It never was. It is a chip race, a metals race, an electricity race, and a logistics race. The visible competition over models is the surface phenomenon. The structure of the next world order is being decided one substation, one foundry, and one processing plant at a time.

Whoever moves first, builds fastest, and locks in the supply chains wins. Whoever waits loses, regardless of how good their software is.

The window to choose which side of that line a country, a company, or a portfolio ends up on is open right now.

It will not stay open for long.

Within hours of that read, the largest aircraft carrier in human history, the USS Gerald R. Ford, was confirmed leaving the Middle East. Brent crude broke above $120 per barrel for the first time since June 2022. Gasoline at the American pump averaged $4.23, the highest since 2022. American wheat futures jumped roughly four percent on Tuesday alone, on top of an already thirty percent year-to-date gain. The same Pakistan that hosted Trump’s failed shuttle delegation last week opened a series of land corridors from Karachi and Gwadar to the Iranian border, with thousands of cargo containers reportedly waiting in port for the routes to clear. The first real workaround to the blockade since the operation began.

The mainstream media is reading all of this as Trump backing down. The harder framing, and the one a non-American watching from outside has had reason to notice, is that Trump just walked into a siege and signaled out loud that he has all the time in the world.

On Truth Social this morning Trump rejected Iran’s latest proposal to end the war and posted that the regime “can’t get their act together” and “better get smart soon.” He retruthed a separate post simply rebranding the Strait of Hormuz as “STRAIT OF TRUMP.” Read together, the posture is precise. The negotiating clock is now Iran’s problem, not Washington’s. The waterway the regime has spent forty years pretending to control is being treated, in writing, as American real estate. The administration is no longer asking for a deal. The administration is waiting for the regime to fail.

The European backstop for the operation is fracturing in public, and most outlets are missing the pattern by reporting each crack separately. Sweden’s prime minister Ulf Kristersson said this week that he cannot see any strategy behind the American war in Iran. Germany’s Chancellor Friedrich Merz called the regime’s negotiating tactics a humiliation of the United States. Per Peter Zeihan’s same-day reporting, Italy has now gone fully silent on the war and stopped logistical support to Israel, and the European piece of the American expeditionary chain is now down to Germany alone. The administration’s response, posted on Truth Social today, was to study reducing American troop levels in Germany. The same disciplinary cycle used on Spain (a NATO suspension review) and the United Kingdom (a Falklands hint) is now extending to Berlin. The countries that did not show up for the war are being individually invoiced.

Inside the American economy, the Federal Reserve admitted defeat in its own quiet way yesterday. The central bank held interest rates unchanged for the third meeting in a row. That part was unremarkable. What was remarkable is that four members of the Federal Open Market Committee dissented from the decision, the first time that has happened since 1992, and the official policy statement upgraded its description of inflation from “somewhat elevated” to “elevated.” The Fed’s own preferred inflation measure is now running at three and a half percent, and Chair Powell’s own statement said energy prices will push it higher in the near term. The odds of any Fed rate cut this year fell to a new low of forty-four percent. The ten-year Treasury yield silently crossed back above 4.40 percent. The Fed is no longer pretending that inflation is somehow temporary. The war is being paid for at the gas pump and the mortgage desk, and the political bill comes due over the next six months as the midterm cycle starts in earnest.

The pattern across all of this is the same. The institutions everyone assumed were permanent are visibly conditional now. The OPEC quota system fell apart over the weekend with the United Arab Emirates walking out. The Iranian regime’s claim to be a functioning state collapsed when its parliament speaker ended up under house arrest and the president went on television asking citizens to use only two light bulbs. The Federal Reserve’s eight-year insistence that inflation was somehow temporary just got rewritten in its own policy statement. The European security guarantee that backstopped Stockholm and Berlin and Madrid since 1949 is now being publicly questioned by Stockholm and Berlin and Madrid themselves.

None of these collapses were created by the Trump administration. The administration is the first government in a generation that named the rot out loud and acted on it in writing, in public, on Truth Social, in a way that forces every other actor to either match the new posture or get repriced.

In the country I come from, the way an old institutional order ends is rarely with one cinematic event. It ends with a sequence of quiet tells, each of which can be denied alone, all of which point the same direction when stacked together.

The siege is the loud story. There are three quieter ones unfolding behind it, and the first one decides who makes the plastic in every Walmart for the next forty years:

Read more