James Aitken on why the price of AI will pop, Kevin Warsh, gold and inflation

ELI5 / TLDR

James Aitken advises some of the world’s biggest investors, and his clients keep asking him when the AI bubble pops. His answer: bubbles don’t pop when everyone’s asking when they’ll pop, so stop trying to call the top and instead figure out how to ride the trend without getting hurt. He thinks AI is the biggest event of our lifetimes, that all the spending on AI and defence and resilience means inflation settles nearer 3% than 2% (and that’s fine, not the end of the world), and that the single most useful thing any investor can do is shut the inbox, ignore the experts, and just watch price — because price tells you what’s true before the story does. Along the way he touches gold’s strange sell-off, the new Fed chairman Kevin Warsh, why the dollar refuses to fall, and the coming wave of mega-IPOs.

The Full Story

Stop asking when the bubble pops

The question Aitken hears most, by a mile, is “when does the AI bubble pop?” His reply is a piece of market folk wisdom that happens to be true:

“Bubbles tend not to pop when everyone’s asking when does the bubble pop.”

His mental model for AI is the early-2000s China property and infrastructure boom — a “colossal secular trend” that drove everything for years. The lesson from that era wasn’t to be clever and pick the top. It was to not overthink it, grasp how big it could get, and stay invested without losing your shirt. He admits he was slow to understand AI and is still, by his own description, “way down the scale of AI adopters” — overawed, not overwhelmed.

He splits his clients into two camps: those outside the AI ecosystem peering in trying to understand it (where he places himself), and a lucky handful sitting at the absolute epicentre — people who funded CoreWeave’s $9 million round in 2019, or who supply gas turbines and fuel cells, or who write the code inside Anthropic. The second group is how he learns what’s actually happening.

Bottlenecks, and why 3% inflation is okay

Aitken’s framing of inflation is refreshingly plain. The 2% target, he points out, isn’t a law of physics — it’s a number central banks more or less made up (the Reserve Bank of New Zealand first, then Greenspan informally, then Bernanke officially) and backfilled with academic justification afterward. The Fed never actually got inflation back to that number after the post-COVID surge.

Now lay on top of that a world spending trillions on the AI arms race, on defence, on “national resilience” — all the things rich democracies neglected for decades while optimising for efficiency. We’ve flipped, in his phrase, from just-in-time to just-in-case. That switch costs money.

“Our future is looking more expensive by the day.”

The key insight: choosing resilience over efficiency is a deliberate trade-off, and it implies we won’t crash the economy back down to 2% just for the sake of the target. So if US inflation runs closer to 3% than 2% for a while, that’s the price of doing things we should have done anyway. As long as policy stays credible, you “muddle through.” Higher inflation also means long-term bond yields clear a bit higher on average — and the world already lived through 10-year Treasury yields above 5% in late 2023 without a recession. (The pain only landed on businesses that had run “way, way, way ahead of their earnings momentum.”)

One open question he flags honestly: maybe all this investment makes the economy less sensitive to high bond yields, so the threshold that triggers a recession is higher than it used to be. Or maybe lower. “We shall find out” is a phrase he uses a lot, and it’s a feature, not a bug.

The market is a casino — use it

Aitken shares Buffett’s worry that markets have turned into a casino: single-stock leveraged ETFs, Polymarket, the retail frenzy. He watches market structure nervously — especially the way passive concentration means a stock that misses earnings by a hair gets “absolutely destroyed” because there’s no liquidity on the other side. If the AI bellwethers ever start doing that, look out.

But there’s a flip side, and this is the practical takeaway. If you understand why a good business falls apart on nothing — an “air pocket” of liquidity rather than a real change in fundamentals — you can step in as the next buyer. Think of it like this: the casino structure occasionally hands you a great company at a panic price, and if you understand the plumbing you collect the margin of safety. He’s not worried about equities while earnings estimates are still rising; a bear market with rising earnings estimates would be “very, very strange.”

Picks and shovels: dig the dirt, lay the cable

True to being “a simple Australian,” Aitken stays in his circle of competence: resources and commodities. If the world is building and powering data centres everywhere, it needs a lot more stuff. The big miners — BHP, Rio, Glencore — are unusually disciplined this cycle, having learned not to top-tick by overbuilding supply. Less supply elasticity plus capital discipline plus dominant position equals strong earnings power. He notes materials are still only ~2% of the S&P 500 — a weighting that “smells like 1999” — and suspects the median fund is underweight, partly because these were the very industries defunded under ESG, which he calls “an act of self-harm.”

Two specific company stories illustrate his “price moves, then the narrative follows” thesis:

Bloom Energy — a “25-year overnight success” in fuel cells. The pitch: while everyone obsesses over how a data centre takes two-to-three years to plug into the grid, Bloom can power one in 55 days. He uses it less as a stock tip than as a metaphor for how fast the energy-for-AI landscape is being disrupted (and notes, with a nod to the interviewer, that Bloom itself may already be getting disrupted — disruption all the way down).
Prysmian, the Italian cable maker — one of his “great misses.” A client flagged it years ago: full order books, brilliant capital allocation, no one paid attention because it’s Italian. The share price sat dead for years, then ran from ~40 to ~150. The price moved first; the story caught up later.

Gold’s strange behaviour

Gold ran well past $5,000 as central banks — especially in countries spooked by the West confiscating Russian reserves — stockpiled it and took physical delivery onshore. Then something odd happened during the Iran turbulence: gold sold off hard, with Asian central banks (Turkey, possibly Poland) selling.

Aitken’s read is that this was perfectly rational, even though it surprised everyone expecting gold to be a crisis hedge:

“What’s the point of having insurance if you’re never going to use it?”

Central banks that had hoarded gold for years simply used it — to pay for pricier imports during the crisis, locking in enormous profits. The deeper principle, which generalises far beyond gold:

“I pay close attention when the narrative around a popular asset starts to change, and more accurately when the price diverges from the narrative.”

If you’re more and more confident in a business and the stock keeps leaking, you’ve missed something — the market is trying to tell you. The next test for gold’s structural story, he says, is what happens if the Fed turns to cutting again or the dollar starts to weaken — neither of which is happening now.

Kevin Warsh and the new Fed

On the incoming Fed chairman: Warsh is ambitious (he’s lobbied for the job for a decade) and has clear views the market broadly likes — less reliance on models, less forward guidance, tighter communication discipline, better debate around the table. The one thing markets care about most isn’t where rates go but what he does with the still-enormous balance sheet.

Aitken pushes back on the lazy take that Warsh is a “patsy” or “sock puppet” for Trump. He points to symbolism: Warsh was introduced in the East Room of the White House — the first time in 20 years a Fed chair got that treatment — and Trump, surrounded by flags, didn’t confirm the market’s fears that Warsh would be a lapdog. A rule of thumb worth keeping:

“When the president stays on script, I pay attention. When the president goes off script, I ignore him.”

His advice to clients: don’t bet against Warsh head-on, but don’t bet on him blindly either — give him the benefit of the doubt, because behind the scenes there’s genuinely impressive, joined-up thinking (Warsh, vice-chair Bowman, Treasury’s Bessent) about the balance sheet and the feedback loops between post-2008 bank regulation and Fed policy — connections the Fed has avoided thinking about systematically since the crisis.

Why won’t the dollar fall? Watch the yen

There are a dozen reasons the dollar “should” be lower, and it isn’t. Aitken’s hard-won lesson from 35 years in FX (starting at Macquarie in Sydney): you cannot hold a view on the dollar without a view on the yen, because Japan is an enormous exporter of capital. And Japanese savers keep buying dollars to invest abroad. When the Bank of Japan intervened to push dollar-yen down from 160, savers simply bought dollars again on the dip, and it drifted back up. This is the behaviour of a regime — under PM Takaichi, whom he admires — that doesn’t want a strong currency, just one that doesn’t collapse.

So the yen is fundamentally cheap, but you don’t express that by buying the yen (it keeps not rallying). You buy yen-denominated stocks — the Nikkei and Topix win precisely because Japan is “getting away with a really weak exchange rate.” And the contrarian risk he’s mindful of: with the Fed now looking more hawkish, the genuine pain trade may be a dollar that doesn’t fall but actually rises.

The coming flood of mega-IPOs

Borrowing a point from Paul Tudor Jones (on Patrick O’Shaughnessy’s podcast): part of why equities have done so well is that buybacks keep shrinking the supply of shares. We think about supply and demand for commodities but rarely for shares themselves. Now two things are about to reverse that:

Supply surges as SpaceX, OpenAI, Anthropic and others come to market — a “colossal unlock” distributing enormous gains to endowments, foundations and early investors. The SpaceX prospectus is being pushed even to Australian retail platforms because they need to place it everywhere.
Buybacks slow because the AI hyperscalers, once self-funding their data-centre build-out from cash flow, are now borrowers.

So the question this summer is less about price and more about who digests all this supply. And — “follow the money” — whatever those endowments get back from the SpaceX windfall, Aitken bets not a dollar goes back into private/illiquid investments; it all flows into public markets. There’s even a dollar angle: the Gulf and other capital owners couldn’t sell their illiquid US holdings because there was nothing to sell — but as IPOs hand them actual dollars, currency markets become a real-time “thumbs up / thumbs down” on the Trump administration.

He also notes the cynical mechanics of index inclusion: every CEO knows getting into the right index unlocks passive flows that don’t care about your business, only that you tick a box. SpaceX reportedly enters at 3x its free float, “guaranteeing that an awful lot of people get into this at a very, very high price, which they may come to regret.”

What the elite investors actually do

The richest part of the conversation isn’t macro at all — it’s about temperament. Asked what the truly great investors do differently, Aitken describes a client of 25 years who quietly harvested billions from a corner of a hedge fund with zero public profile. His routine, every single day for 35 years:

“I will never read any emails in the morning. I will never read any broker reports. All I look at… I will just look at price.”

Why? Because price is how you keep score — it’s the P&L, the NAV, how you get paid. Most days nothing’s happening and prices are roughly where you’d guess. But occasionally something moves that no one is talking about, and that’s the signal. Even more striking: before putting on a single unit of risk, this man checks himself — Have I slept? Am I in balance? — because he knows that if he’s right he’ll be running it hard, so he wants to be sure he knows his own state of mind first. And when he can’t sleep at max risk, he just sells everything; the Soros “my back hurts” instinct made into a discipline. Aitken has started keeping a feelings diary himself (and is drinking marginally less red wine).

The framing that ties it together:

“So many people approach investing as ‘I am right.’ These wonderful people start with ‘How might I be right?’”

The great ones are, in his words, “wonderfully weird” — comfortable being unconventional, brilliant at pattern recognition, and that comfort is the edge. The general principle he keeps returning to: don’t read another research report; lean into what your best, longest-held portfolio companies are telling you, look for patterns across them, and protect the most precious asset — time to think.

China, and the three-year bet

China barely registers in his client conversations now — not because they’re bearish, but because there’s so much going on elsewhere. He thinks Xi is deliberately trying to build an equity culture (copying Japan and Korea’s success), and that Jack Ma’s rehabilitation was the real “starting gun” — “you didn’t need to read another China research report.” Trump’s recent Beijing visit he dismisses as “one of the great nothing burgers” — pure ego-management, by design. The one thing worth watching: Chinese industrial companies are raising prices, which could mean the great disinflation that China exported to the world is fading — a subtle but important shift for Western inflation.

Asked what we’ll realise in three years that we don’t understand today, his answer is unhesitating: that AI was the biggest event of our lifetimes. He leans on Harvard decision theorist Richard Zeckhauser (80, an advisor to Munger, “not prone to overstatement”) saying flatly that AI is the biggest event of his lifetime. Only ~0.1% of the world’s population, Aitken figures, currently understands what something like Claude can do. What happens when it reaches 5%?

Key Takeaways

The crowd’s favourite question is a contrarian tell. When everyone’s asking “when does the bubble pop?”, it usually isn’t about to. Bubbles top in euphoria, not anxiety.
2% inflation is an invented target, not a natural constant — first the RBNZ, then Greenspan informally, then Bernanke officially, with academic justification backfilled afterward.
Resilience-over-efficiency is structurally inflationary. The civilizational switch from just-in-time to just-in-case (defence, AI capex, reshoring) means inflation likely settles nearer 3% than 2%, and that’s an acceptable trade-off, not a catastrophe.
Higher bond yields haven’t broken anything except businesses that ran far ahead of their earnings. The 2023 episode of 5%+ Treasury yields produced no recession.
Passive concentration is a structural fragility: thin liquidity means a small earnings miss can devastate a stock — but it also hands disciplined buyers great businesses at panic prices.
Price moves first, narrative follows (Prysmian: dead for years, then 40→150). Watch for price diverging from narrative — it means you’ve missed something.
Gold as insurance you actually use: central banks selling gold during the Iran crisis to fund imports wasn’t a failure of the gold thesis — it was the insurance working exactly as designed, at a profit.
You can’t have a dollar view without a yen view. Japan’s capital exports dominate the dollar complex; Japanese savers keep buying dollars on dips, so the cheap yen stays cheap. Express the bullish-Japan view via Nikkei/Topix, not the currency.
Buybacks are stealth demand; mega-IPOs are stealth supply. Hyperscalers shifting from cash-flow buyers to borrowers slows buybacks just as SpaceX/OpenAI/Anthropic flood the market — the summer test is who absorbs the supply.
Index inclusion is a re-rating game. Passive flows care that you tick the box, not what your business does; IPOs are sized to clear inclusion thresholds.
The elite-investor discipline: look at price before reading anything, and check your own state of mind before putting on risk. “How might I be right?” beats “I am right.”
Watch Chinese industrial prices. If China stops exporting disinflation, Western inflation gets harder to tame.

Claude’s Take

This is a strong conversation, and the score reflects it. Aitken is genuinely good — he thinks in second-order terms, he’s honest about what he’s missed (AI, Prysmian), and he resists the easy guru move of issuing confident predictions. “We shall find out” is his refrain, and it’s a sign of someone who’s been humbled by markets enough times to mean it. The framing devices are clean: insurance you actually use, just-in-time to just-in-case, price diverges from narrative, “how might I be right.” These are portable mental models, not just hot takes.

Where to apply the BS filter: there’s a fair amount of name-dropping (the unnamed billionaire client, the South-of-France birthday, Zeckhauser, Paul Tudor Jones) that functions partly as credentialing. It mostly earns its keep — the anonymous-trader portrait is the best part of the episode — but you should notice that “an investor I know who’s at the epicentre says X” is an appeal to authority dressed as analysis. The Bloom Energy and Prysmian stories are vivid but, by his own admission, partly retrofitted (“the great miss”); survivorship bias lurks in any “the price moved and then the narrative followed” anecdote, because you never hear about the ones where price moved and the business was just broken.

His inflation-runs-at-3%-and-that’s-fine thesis is reasonable and increasingly consensus, which is worth knowing — it’s no longer a contrarian call. The most useful, least hedged, and most durable material is the temperament section: look at price first, check yourself before sizing risk, “how might I be right?” That’s the part worth re-reading, and it’s why this lands at 8 rather than 7. Two reasonable people could dock it a point for the conversational sprawl and the occasional glide past hard questions (the dispersion puzzle gets a candid “I don’t know,” to his credit), but the signal-to-noise is high for a 90-minute macro chat.