The US 10-year Treasury yield reached 5% on the morning of this interview, its highest since 2023, and David Woo says it only has to reach 5.25% or 5.30% for the artificial-intelligence trade to come apart.
The usual reading of a 5% yield is a fiscal one — too much government debt, too few buyers. Woo's is that the marginal borrower is no longer the Treasury but the companies building data centers, and that the administration is now trying to hold long-term rates down to protect them.
"There's no doubt that surging bond yields is largely the result of this out of control AI capex."
Woo ran global rates, foreign exchange and emerging markets at Bank of America and global foreign-exchange strategy at Barclays before setting up his own research firm, and he came into this conversation short stocks and long oil.
The full interview is covered here so you can skip it. 44 minutes of audio, 25 minutes of reading.
Here are the 18 calls that matter.
👤 Guest: David Woo, founder and CEO of David Woo Unbound, previously Head of Global Rates, FX and Emerging Markets at Bank of America and Head of Global FX Strategy at Barclays, with a PhD in economics from Columbia
🎙️ Host: David Lin, a former BCA Research macroeconomics researcher who now runs The David Lin Report
📰 Published: 15 September 2026 on YouTube (David Lin) · recorded 14 September 2026
🔴 YouTube | 🟢 Spotify | 🟣 Apple Podcasts | ⏱️ 44 min | ✅ Time saved: 19 min
Key Takeaways
Oil and bond yields will keep rising until equities fall, because nothing else stops either one
His estimate of what it takes to change the policy response is a 10% drop in the stock market
The level that breaks the artificial-intelligence trade is 5.25% to 5.30% on the 10-year, not 6%
The bond market is repricing because of corporate capital spending, not government deficits
The hyperscalers are in negative free cash flow and have committed at least $3 trillion of off-balance-sheet spending
Anthropic's own shareholders are selling into the IPO rather than holding for it
The company is seeking $2 trillion; he last saw the secondary market nearer $1.3 trillion
He expects Trump to ban Chinese artificial-intelligence models, and expects China to answer with rare earths
Chinese models are already at 25% share in the US on his figure, and he says 50% within six months
China caught up on worse chips than it had 18 months ago, which is why he says nobody still believes the gap is months
The Fed hiking into this looks to him like 1987, when a three-month-old chairman was building credibility
Being bearish is not the same as being short all the time — he is up 3% against a three-month Treasury bill after a year of shorting stocks
1. Short Stocks, Long Oil
Lin opened with the two prices that had moved: oil firmly above $100, with Brent near $108 and WTI near $103, and the 10-year Treasury yield at 5% for the first time since 2023. Equities were selling off as they spoke. He asked whether it was the oil or the yields.
Woo said both, and that the surprise was how little the stock market had given back. "I'm short stocks and I'm long oil and I'm surprised the stock market isn't down much more today." He was equally surprised by the previous two weeks, in which yields and oil had both surged and equities had not.
He put the puzzle as a question about what equity investors think they own: "Is it thinking like, the stocks have become the new safe haven?"
His framework is a loop with only one exit. "So oil prices is going to keep going higher and yields, I can tell you there's no doubt that bond yields are going to keep going higher until the stock market goes down."
The mechanism on the oil side is political: he said Trump will not reverse course — Woo's word for it throughout is "taco" — until equities fall.
The mechanism on the rates side is a central-bank reaction function. Woo said central banks read a rising stock market as easing financial conditions, which leaves them comfortable letting yields rise further.
His conclusion was that equities are nowhere near out of the woods.
2. Oil Is the Chicken
Lin asked which was the bigger risk to equities — the Iran war pushing oil higher, or the Fed raising rates, noting that the 2022 hiking cycle was bad for risk assets and bonds alike. The CME's gauge was pricing a 90% chance of a hike at the Wednesday meeting.
Woo put oil first in the causal chain. He said there is no doubt the Fed is poised to hike because oil prices have gone back up, and that if oil collapsed tomorrow the rates market would follow — "rates will probably tumble fairly quickly after that."
He relayed ECB President Christine Lagarde saying the week before that the longer oil prices stay up, the greater the risk of a "second round effect into inflation".
The repricing is not a US story. Over the previous six weeks, he said, the rate outlook has been repriced for every major central bank — the Fed, the Bank of England, the Bank of Canada and the Reserve Bank of Australia.
The ECB has already moved. It hiked once the week before, and on his account is now priced for at least one more before year end, possibly two.
His name for what this adds up to is a synchronized monetary tightening caused by an oil shock rather than by a demand boom.
3. A Call Spread on WTI
Asked how high oil goes, Woo described the position rather than giving a price target, and said there is more than one way to trade the view.
He owns a call spread on the December WTI futures contract, struck at 80.95 on the lower leg and 105 on the upper, expiring the day after the midterm election. Oil had just traded above the lower strike as they spoke.
The structure, not the forecast, is the point: he said a move to 105 pays about ten times the premium he put up, and that this asymmetry is what an option buyer should be hunting for. He added that he knows oil can go to 150, but that he has been there before and expects Trump to intervene, so he set the target conservatively.
His timing argument is electoral. "The Iranians know that before the midterm election, Trump is politically constrained and therefore they have a much better chance to basically force him to taco" — that is, to return to the memorandum of understanding Trump signed in July.
The second leg is the Houthis, who on his account "have not been involved in this war" until August. They were not there in March, April, May, June or July. He attributes the change partly to the Saudis overplaying their hand.
Their entry doubles the number of chokepoints Iran can squeeze. "Iran now can throttle both the Strait of Hormuz as well as the Red Sea."
He pointed to a physical crude spread in Europe that had blown out to its highest level since April as evidence that tanker traffic through the Red Sea into Europe has fallen sharply since the Houthi attacks began, and said the tightness is now showing up in Asia as well.
4. Asia's Barrels Are Thin
Lin put the bull case's main objection to him: China alone reportedly holds more than a billion barrels of reserves, so the question is not whether the conflict escalates but whether importing countries can ride it out.
Woo's answer was that holding reserves and being willing to spend them are different things. "The fact that they have a billion barrels doesn't mean that they're going to want to dip into it much more than they already have."
The Chinese import data has turned. Imports fell in March, April, May and June; July was up on June and August up again on July. "Chinese oil imports have gone up now two months in a row." He noted the level is still below where it was before the war.
He cited anecdotal reports of Chinese teapot refiners shopping for crude around the world, and said the Oman futures contract against Brent — his proxy for what Asian refiners pay versus European ones — shows a rising premium.
On the United States he said the usable part of the strategic reserve is much smaller than the headline. "And in the case of the US as you probably know the operational limit in terms of strategic reserve is probably closer to 150 million maybe max maybe 200 even and we're already at 280."
Japan has gone the other way. Two weeks before the interview, he said, the government instructed refiners that no strategic reserve would be released into the market in September or October, which is why Japanese refiners have been scrambling to buy.
His summary of the region was that the buffer is not there. "The Asian market inventory is actually pretty depressed and the governments are not willing to underwrite this physical shortage indefinitely."
5. What Takes Oil Past $120
Lin said traders were pricing a 35% chance of oil above $120 and asked what would get it there.
Woo's answer was that very little would. "I think it won't take that much actually. It really won't take very much."
The story he says everyone missed is why oil traded in the $80s through August. Tankers owned by Saudi Arabia and the UAE were moving crude out of the Strait of Hormuz and passing it to other tankers on the other side, bound for China and elsewhere — and US officials, he said, read that as proof the Strait was open.
His reading is that the traffic existed at Iran's discretion. Iran was deep in negotiation with Oman — representing Saudi Arabia, Kuwait and others — over a deal that would give Iran control of the Strait, and "Iran didn't want to blow up Saudi tankers in the Strait of Hormuz" while that was live.
That negotiation is now falling apart, which he said is why the meeting was cancelled the day before, and why Iran is having the Houthis strike Saudi Arabia directly to force a signature.
He said Washington has been working against the deal, and cited Trump saying three weeks earlier that he wanted to bomb Oman.
The consequence if it collapses is binary. "If this deal falls apart, if Saudi because of US pressure refuses to sign onto this deal, then Iran said screw it. They're going to basically not allow a single tanker to get through and the oil price could easily go up."
6. The Yen and the Rupee
Lin, noting Woo's time at the IMF in the 1990s, asked whether a sustained oil shortage sets up another Asian financial crisis.
Woo declined the parallel. "The Asian crisis in the '90s, it was not about oil. It was about leverage. It was about fixed exchange rate regime." Mismatched liabilities were the third leg.
Pushed on the simpler point — that a country importing oil through a shortage should see its currency fall — he agreed, while noting oil is a smaller share of Asian imports now than it was then.
Japan is where he thinks it bites. "The Japanese yen is already the weakest currency in the world," and both the central bank and US Treasury Secretary Scott Bessent are trying to hold it up.
The reason a US Treasury Secretary cares about the yen is the chain that runs from it: "And Bessent is trying to shore up because he's worried that a run on the yen is going to basically lead to the mother of all sell off in US Treasuries." Higher oil weakens the yen, and a weaker yen ends up in higher Treasury yields.
India is the other large importer he named, and the one he says is already paying. It is buying Russian and possibly Iranian crude, and its stock market has been among "the worst performing stock market this year", which he put down to oil's effect on inflation and the current account, with the rupee very weak alongside it.
Lin's own observation cut against the sequencing: the yen has restrengthened since intervention, and the 10-year still reached 5% without a yen collapse or a foreign holder dumping Treasuries en masse.
7. Bessent's Losing War
Asked what Bessent does now that the 10-year is at 5%, Woo said the Treasury Secretary had overplayed his hand, and then gave the reason he thinks the Treasury moved at all.
He dated the decision to cap long-term rates to a single day about a month earlier, and said its timing was not a coincidence: the same day brought news that Anthropic was preparing a $2 trillion IPO and that OpenAI had announced a $500 billion data-center project with $125 billion from Nvidia.
The connection he draws is direct. "There's no doubt that surging bond yields is largely the result of this out of control AI capex."
The mechanism is a funding gap. "The hyperscalers are mostly in negative free cash flow," and their earnings are growing more slowly than their capital spending, so every additional dollar of capex is an additional dollar of debt issuance.
On top of that sit the commitments already made: "That is to say that there are at least three trillion dollars of debt coming through" the pipeline, from more than $3 trillion of off-balance-sheet commitments. "This is what the bond market is rebelling against."
So the Treasury's objective, in his reading, is not fiscal at all — it is to stop surging bond yields "choking off the AI bubble."
He thinks the tool is far too small for the job. "You cannot fix this problem with eight billion dollars of whatever Treasury buyback."
The buyer base is the other half of the problem. "China is selling US government bonds every single month," and he said the selling accelerated over the previous three months, leaving Japan as the largest foreign holder.
Japan is now leaning the same way. He said the Japanese government met the country's pension funds the previous week to say the yen is too weak and there are too many JGBs, and to ask them to consider selling Treasuries and bringing the money home.
Norway's central bank, which runs the largest sovereign wealth fund in the world, has said it will cut its holdings of US government bonds by $80 billion.
His verdict on the policy is that the administration is willing to engage in "financial repression" to save the artificial-intelligence bubble.
8. Crowding Out
Lin restated the thesis as a conditional — if rates go higher, the artificial-intelligence bubble pops, and that is what the Treasury is trying to prevent. Woo agreed and gave the textbook name for it.
The supply of savings is not growing. Government saving is getting worse, and "Household saving is very very low as only at 3%."
Against that, investment is surging, which pushes long-term rates up — and higher long-term rates "crowds out everything else from mortgages, home construction", and possibly even artificial intelligence itself.
The evidence he offered was the summer's price action. In July, he said, the Nasdaq was very negatively correlated with bond yields: "When bond yields went up, Nasdaq went down."
He then closed the loop between the two markets and the political constraint. "I think the stock market have to fall 10%. For Trump to taco for oil price to stop going up." If equities do not fall, Microsoft and Amazon keep issuing debt to fund capital spending and rates keep climbing until they do.
9. His Equity Shorts
Asked to confirm the positioning, Woo gave the structure.
He is short equities through options rather than outright: "I'm long put spreads on QQQ, and then I'm long on out of money basically put on spot."
Both expire before the midterm elections, because that is the window he thinks is live. He said the period between the interview and the midterms is when things get most interesting again.
10. Bomb the Data Centers
Lin asked whether the artificial-intelligence bubble could pop on its own, without help from rates or oil. Woo said yes, and spent the rest of the interview on what would do it.
He started with the essay Dario Amodei published the Saturday before, which he said he had been making the same argument about for weeks. The passage he singled out was not the call to slow down but the condition attached to it: "However, he said the pace of the slowdown should be limited so that China does not basically overtake us in the process."
The second data point was a conference remark ten days earlier by Jacob Stokes, a former Obama administration official now at a think tank. On Woo's account, Stokes said that if the US had any evidence China was about to reach artificial general intelligence, it should resort to espionage, to cyberattack, or in extreme circumstances "bomb the Chinese data centers".
Asked by Lin why this is a national-security question at all, Woo pointed to Trump's own framing from the day before — that whoever wins the AI war wins the war, period — and to the range of things artificial general intelligence could be used for.
11. Why Astra Was Allowed
Woo then laid out a three-week sequence he says only makes sense as a policy decision.
Three weeks before the interview, OpenAI said it had paused its work on Astra because of growing security risk. Two weeks before, a hundred companies warned that cyberattacks were becoming more sophisticated and more widespread. A week before, OpenAI launched Astra and called it the beginning of artificial general intelligence.
Woo's explanation of the reversal is that the government chose between two risks. "The Trump administration's decision to allow Astra to be launched was because the Trump administration decided that the risk of allowing China to reach AGI was greater than basically a massive AI enabled basically cyber attack."
The reason the government gets a vote is his broader claim about what the technology now is: "Because AI, this technology is no longer commercial technology, has become a battlefield for national security for the race for power."
12. Banning Chinese Models
The policy conclusion he draws from all of it is a trade ban, and he thinks the groundwork is already laid.
He pointed to the FBI and CIA saying the previous week that China has been engaged in industrial-scale distillation of US models. "This statement is extremely important because it provides Trump with an excuse to ban Chinese AI models," which he expects.
The market-share numbers are what he says forces it. "Chinese AI now has a 25% market share in the US." And: "I can guarantee you, David, in six months time it will have 50%."
He read Amodei's essay as reinforcing the same conclusion, because slowing US development only works if China is prevented from taking the share that a slowdown frees up.
Lin asked whether a ban lands before the Xi Jinping meeting on 24 September. Woo said he is not even sure the meeting happens, while the market assumes it does.
The retaliation is the part he thinks is underpriced. If Trump bans Chinese models, "China very likely will basically retaliate by essentially playing their rare-earth card. And that will be very bad news for the entire industrial complex, and everything else."
Lin agreed it would be a major escalation, and noted the asymmetry running the other way: artificial intelligence is not yet a large component of the Chinese economy, while a rare-earth export ban would shut down whole sectors of the US economy.
13. The Truth Social Post
Lin put Trump's post on screen and read it out: the only control or guardrails artificial intelligence needs is a strong, smart, high-IQ president; a swipe at Amodei for "now pretending to be a perfect little angel"; and a claim of "tremendous criminal and regulatory power over these companies". The line Lin asked about was "There is a sick conspiracy going on against AI and data centers. Whoever wins AI wins."
Woo's first response was about the situation rather than the post. "We're at a point here now where we're talking on a YouTube channel trying to decipher what the president is saying on Truth Social. This is what our financial community has come to."
He then gave the arithmetic he says explains the administration's behavior: the US is about 15% of world GDP, and its stock market is 60% of the world's stock market, with the Magnificent 7 at about 30% of that and technology around 40%.
"Right now the AI valuation the AI trade is the US economy right now." Without it, he said, "If it hadn't been for the AI capex, the AI bubble, the AI trade, the US stock market would have already collapsed" under rising oil and long-term rates.
That is why he does not expect the administration to blink. It "will not stop at anything" to protect the artificial-intelligence trade, because if "anything happens to this thing, it's game over."
"Trump has thrown everything behind this AI trade. He has basically unleashed the entire federal government to underwrite this entire thing, this boom." Woo called it arguably his greatest achievement, and said that is exactly why a failure would be terminal.
He also said he does not want to bet against Trump, citing reports in Politico and The Hill that Trump has stopped listening to aides, senior officials and congressional Republicans — which to him raises the odds that the ban happens.
14. China Already Caught Up
Lin said the consensus among the economists and engineers he talks to is that Chinese models are five to ten months behind American ones, and asked whether they eventually catch up. Woo rejected the premise.
"I don't believe anybody thinks the Chinese are 5 10 months behind the US anymore." He called it the most unexpected development of the year and said he did not see it coming.
Of the top-ranked models today, he said, about half of the top six are Chinese, at performance levels comparable to their American counterparts.
The part he thinks matters is what they did it on. Chinese labs have no access to the most advanced chips — not Rubin, not even Blackwell. "They're still using the H800 chips."
The gap has widened rather than narrowed since DeepSeek's first release 18 months ago, when DeepSeek was on H800s and OpenAI on A100s: "If you look at the differential in terms of chip performance between what US companies have access to versus what the Chinese companies have access to, the differential today is much wider than it was 18 months ago."
It is also no longer one company. Where DeepSeek was on its own, he counted five or six Chinese companies racing now. His phrase for the state of play was that "the cat is out of the bag".
This is the reason he says the US cannot afford to slow down even at the risk of a systemic cyberattack, and he called it "the biggest single tail risk" for the entire global stock market.
15. $2T Ask, $1.3T Bid
Asked whether he would buy the Anthropic or OpenAI IPOs, Woo answered on valuation and then produced the number he finds most telling.
"There has to be a price for everything not at two trillion dollars."
There is a private secondary market in Anthropic shares, and it is not agreeing with the ask. Against a $2 trillion raise, "The shares, if I recall, the last time I had looked, was trading more like $1.3 trillion."
His point is about who is selling. An insider could hold for six weeks and cash out at the IPO price; instead they are selling at 1.3.
The control for that behavior is the last comparable listing: "6 weeks before the SpaceX IPO, there was nobody selling because everybody's sure they were going to make a lot of money."
He also read Amodei's essay through the same lens. He said the roadshow would have started the week before if the listing is in October, and that every investor meeting would have carried the same questions about China and cybersecurity — so the essay is the answer to those questions rather than an unprompted act of conscience. "You believe that guy's paid? That guy's the most greedy person I know."
The listing itself is the event he says the whole complex hangs on. "The Anthropic IPO is extremely important because for Trump like it's a referendum of the entire AI trade." "If it goes poorly, the whole AI trade could basically collapse and then it's the end."
He noted Nvidia is said to be interested in putting $10 billion in as a lead investor to anchor the deal, and said banning Chinese models before the IPO makes short-term sense on exactly that logic.
16. The Shortage Is Ordered
Lin put the counter-case from the sell side: one analyst had told CNBC that demand for inference still far outstrips supply, which is bullish for chips even if the model companies are expensive.
Woo's first response was about what is already in the price rather than about the demand itself.
The growth has a cause that cuts the other way. It is growing so fast because "prices are coming down very rapidly", and that is "the problem for all these companies" — a margin problem for OpenAI and Anthropic, not only a volume story.
The order book may not mean what it appears to. Companies are ordering six months or a year ahead because everyone believes there is a shortage, and those orders "may not be even binding orders." His conclusion: if there is a sudden pullback, "everybody will be pulling back."
He then set out everything the administration is willing to do to keep the boom going — cap long-term rates, bring the federal government in against state-level attempts to legislate against data centers, ban Chinese models.
The trade he takes from that is the trade-off itself. Keeping it going costs either higher rates or a ban on Chinese models, and neither is good for equities: "Trump cannot have his cake and eat it too and that's the way I'm trading this."
17. 5.25% Breaks It
Lin came back to the question he opened the show with: how high do rates have to go before things break, with the 10-year already at 5%.
"I don't know. I think 5.25, 5.30. I don't think it's going to be like 6% or anything like that by the way." Lin's reaction was that the market is almost there.
Woo then added the Fed's own contribution, and it is a personnel argument. "Kevin Warsh really screwed up at Jackson Hole."
His account of why: "He thought that the bond market wanted him to be hawkish because he was told by all these strategists, investors that rates are surging because the lack of credibility of Kevin Warsh." So Warsh set out a black-and-white framework — if core PCE stays above about 2% for long enough, rates have to rise — which Woo says lacks nuance and licenses the market to price more hikes. His verdict: "he is pouring fat to the fire."
The 1987 parallel is about the newness of the chairman rather than the level of rates. "The 1987 stock market crash was preceded by Alan Greenspan becoming chairman of the Federal Reserve only 3 months earlier." A new chairman facing things getting out of control tries to build credibility.
And the crash came on the expectation of the second hike, not the first: "All I can tell you is the Fed raised rates once and then when the market thought they going to raise rates again, that's when the stock market crashed in '87." He allowed that every situation is different.
Stacked on top he named the Fed credibility story, the Japan story and the European story, and said there is a lot he does not like about the world economy right now.
His expectation for the week's meeting inverts the usual logic of a hike. "If the Fed were to hike rates this week, rates are not going to go down. It's going to go up again."
18. Bearish All Year, Up 3%
Lin's last question was the one he says he gets in the comments constantly: the bear case has been made for a long time, the market keeps making highs, and the people making it have missed the move. What does Woo say to that?
He answered with his mandate first. He runs a total return portfolio inside an institutional business advising sovereign wealth funds and hedge funds.
"I've been bearish on the stock market all year actually, but thank god I've been bullish on oil."
The scorecard is a modest one and he said so plainly. "My benchmark is three month Treasury bill. So, I'm like up 3% so far."
The lesson he drew is about execution rather than about the view. "You got to go in and out." Shorting and staying short, he said, risks losing your house before the position ever makes money.
The worked example was the summer: "I caught the sell off in July very well and I got out and then I started again in August."
He described this as one of the most difficult years most macro investors can remember, and said that being up on the year while trying to short the market all year is itself the evidence that the going-in-and-out works.
The trigger he is watching is the Xi summit. "The trigger I'm looking for is the whole Jinping summit. Will it happen? Would it not happen? The market right now is pricing it will happen." And: "If I'm right that Trump goes ahead and ban Chinese AI, I think that we're going to be looking at a much bigger sell-off in the stock market."
Bonus Insights
Lin's own framing set up the interview's premise before Woo said a word — that Trump and Bessent must do everything they can to cap the long end of the curve or the artificial-intelligence trade blows up, and that the entire US stock market is dependent on that one trade.
The show's own research carried several of the numbers the conversation ran on: the 90% probability of a hike at the Wednesday meeting, the 35% probability traders were putting on oil above $120, and Trump's Truth Social post, which Lin put on screen.
Woo's vocabulary is worth knowing to read him: "taco" is his shorthand for Trump reversing a policy under market pressure, and it appears in his oil view, his rates view and his equity view as the same single variable.
On the Asian financial crisis, Woo was notably unwilling to take the analogy Lin offered him, saying twice that he did not know whether it held before explaining why the causes were different.
He was explicit that his oil target is set below his own view of the upside. He believes oil can reach 150 but structured the trade to 105, because he expects political intervention before the higher number is reached.
On the Anthropic IPO he distinguished between two questions that are usually merged: whether the technology is real, and whether the listing prices well. His claim is that the second one now decides the first for the market.
Woo's bottom line is that oil and long-term rates will both keep rising until equities fall far enough to force a political reversal, that the level where the artificial-intelligence trade breaks is 5.25% to 5.30% on the 10-year rather than anything like 6%, and that the bigger risk is not the rates at all but a US ban on Chinese models answered by a Chinese rare-earth embargo.
Products, Companies & Tools Mentioned
Anthropic (Preparing a $2 trillion IPO; its secondary market was nearer $1.3 trillion when he last looked, and he calls the listing a referendum on the entire artificial-intelligence trade)
OpenAI (Announced a $500 billion data-center project with $125 billion from Nvidia; paused Astra over security risk three weeks before launching it)
Nvidia (The $125 billion partner in OpenAI's data-center project, and said to be interested in $10 billion as a lead investor anchoring Anthropic's IPO)
Microsoft and Amazon (His examples of hyperscalers that keep issuing debt to fund capital spending as long as equities hold up)
DeepSeek (The 18-month-old benchmark for the chip gap: it launched on H800s while OpenAI had A100s, and he says the hardware differential is wider now)
Invesco QQQ (The vehicle for his equity shorts — long put spreads expiring before the midterm elections)
Norges Bank (The Norwegian central bank, which runs the largest sovereign wealth fund in the world and has said it will cut US government bond holdings by $80 billion)
Bank of Canada, Bank of England, European Central Bank and Reserve Bank of Australia (The central banks he says have all been repriced hawkishly in six weeks; the ECB hiked once the week before and is priced for at least one more)
Politico and The Hill (The outlets he cites for reports that Trump has stopped listening to aides and congressional Republicans)
Books & Resources Mentioned
Dario Amodei's essay on slowing artificial-intelligence development (Published the Saturday before the interview; Woo's reading is that the qualifier — the slowdown must not let China overtake the US — is the operative part)
The FBI and CIA statement on Chinese distillation of US models (Issued the week before; Woo says it gives Trump the pretext to ban Chinese models)
Trump's Truth Social post on artificial intelligence (Read out on air: guardrails are "strong and smart, high IQ president", a swipe at Amodei, and "Whoever wins AI wins")
Watch the full episode:
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: