Jim Grant on the AI Bubble, Decadent Finance, and the Lessons of History

Key ideas

AI capital expenditure race mirrors 1990s fibre optics — $382B capex from the Magnificent 7 in 2025, triple the 2023 level; the demand gap (college students who stop using AI in spring are the biggest identified user group) is the tell; Grant expects a crash before the productivity revolution arrives.
Decadent finance — Fed intervention prolonging cycles means bad actors go unpunished and credit no longer functions as trust between people; “Mr. Market is the best disciplinarian” and his absence corrupts the system. See Decadent Finance.
Private equity in structural trouble — capitalised for near-zero rates, now facing 8–12% debt costs; assets carried at unreasonable valuations; “democratisation” to retail investors is desperation, not benevolence; Nate Kobak: “maximum aggression at periods of maximum risk.”
Röpke’s inflation diagnosis — Wilhelm Röpke’s 1950s formulation: inflation as “a managerial disease of the national economy,” the result of excess claims on production; under paper money, lost purchasing power is never recovered.
The S&P at 40× CAPE — second highest reading in US market history after the 1999 dotcom peak (44.2×); “the theatre of the major financial market top is on the stage” — all the props are present, but the timing is unknowable.

Background — Grant’s Interest Rate Observer

Jim Grant founded Grant’s Interest Rate Observer in 1983. Now in its fifth decade, it costs roughly $2,000 a year and is read by institutional investors, hedge fund managers, and endowment officers. Nassim Taleb: “Jim Grant thinks outside the box. Please read him, listen to him.” David Swensen, who ran Yale’s endowment, called it “on the must-read list of every serious student of markets.”

Grant’s reputation rests on early calls: warning in 1999 that America was “dangling by a thread” at the dotcom peak; warning about mortgage securities before the 2008 crisis; warning that post-GFC Fed policy would eventually produce inflation (materialised 2021–22).

The episode is recorded in late October 2025, the day after Grant’s annual conference at the Plaza Hotel in New York — attended by many of the most sophisticated credit investors in the world.

The AI check-writing contest

The Magnificent 7 (Amazon, Microsoft, Alphabet, Meta, Oracle, and CoreWeave) will spend $382B in capex in 2025 — up 50% from 2024 and triple 2023 levels. These firms now account for 31% of S&P 500 total capital spending, up from 19% in 2019. OpenAI and Anthropic have raised billions every few months and are valued at hundreds of billions.

Grant’s framing: this is not primarily about the technology but about the human response to transformative technology. The pattern: (1) a genuinely marvellous technology appears; (2) investors race to fund it without asking whether demand will absorb supply; (3) overcapacity develops; (4) the crash precedes the realisation of the technology’s promise, not the other way around. The railroad boom of the 1870s ended in the Panic of 1873. The fibre-optics race of the late 1990s produced vast overcapacity; demand did not materialise on the required schedule. “This reminds me a lot of the fibre optics check-writing contest of the late 1990s.”

The demand gap is the diagnostic signal. David Rosenthal (Nvidia employee number four, gifted computer scientist, PhD in electrical engineering) presented data showing that college students are the largest identified sustained user group — and that their demand falls off measurably in spring when academic term ends. “Who else has such a persistent need for plagiarism?”

Grant is explicit that he risks sounding like the person who said “I’ll get a horse” in 1903. He is not denying the technology’s ultimate value. He is arguing that the realisation of the promise follows, not precedes, the crash.

Voltaire/Mark Twain variant noted: Pierre Lassonde quoted a formulation attributed to Voltaire: “history never repeats itself; man always does.” Grant and Green both note, after checking, that most of these attributed quotes probably weren’t said by whom they’re attributed to.

Private equity: democratisation as desperation

Private equity was capitalised for an era of near-zero interest rates. Debt at 3% rather than 8–12% makes an enormous difference to the returns and solvency of leveraged buyout portfolios. Post-2021 rate normalisation means: companies that were viable at 3% debt service are struggling or failing at 8–12%. Over 20,000 private equity vehicles exist globally, most capitalised for the old regime.

Consequence: assets cannot be sold at their carried values; distributions to investors have not arrived as expected; institutions — colleges, museums, endowments — that budgeted for annual draws from their private equity allocations are not receiving them. They are selling positions in secondary markets (like selling a used car) to meet obligations.

The solution private equity found: retail investors. Marketed as “democratisation” — giving ordinary people access to the superior returns of an asset class previously available only to endowments. Nate Kobak’s counter-framing: this is not benevolence but desperation. The sophisticated institutional market is saturated and returning capital slowly; retail is the new liquidity source. “In finance, whenever you hear the word democratising, hide your wallet.” The Chanel-to-Walmart analogy: selling a product that was originally positioned as exclusive to a mass-market audience at a time when the original buyers are dissatisfied.

Kobak’s historical observation: private equity firms have shown “maximum aggression at periods of maximum risk” — they are not systematically smarter than the market; they are momentum-chasers who ran a disciplined strategy in a tailwind.

Grant’s related point: the “smooth return line” of private equity is not non-volatility; it is non-observation. Assets are marked by the fund manager, not by a continuous public market. The smoothness is an accounting artefact, not a real-world property.

Decadent finance

Grant’s phrase for the current regime — derived from the observation that the Fed has repeatedly intervened to prevent corrections, with the effect that bad actors go unpunished. “Markets, when allowed to function properly, go down as well as up. The down portion serves any number of functions. One of which is to skim the bad actors off the stage.”

When the Fed prolongs every cycle — “lends its force, its arm and strength to prolonging cycles and forestalling bear markets” — bad conduct goes uncorrected. A bankruptcy or liquidation “disbarred” the financial equivalent of a lawyer who committed fraud. Fed intervention kept those people in markets where they continued to cause damage. The longer the regime persisted, the further the system drifted from its corrective mechanism.

The phrase “decadent finance” also captures the quality degradation in credit relationships. In a healthy credit market, “credit is man’s confidence in man” — the willingness to lend is a statement of trust in the borrower’s character and capacity. In the current period, that trust has been replaced by documentation, legal structure, and the hope that your lawyer is more clever than their lawyer. See Decadent Finance.

The LTCM case illustrates the limits of institutionalised intelligence: the fund was run by Nobel laureates in economics, with the most sophisticated models in finance. When it failed in 1998, Goldman Sachs traders were trying to value its assets. Emanuel Derman was on the call — and was startled by the depth of the LTCM principals’ sophistication compared with the Goldman traders trying to price their own rescue. The lesson: sheer mental power is neither sufficient nor a reliable substitute for prudence. The same argument applies to the Fed’s hundreds of PhDs who failed to see that $1,000 stimulus cheques + QE + zero rates would produce inflation.

The fiscal position

Grant cites the statistic: it took the US 222 years to borrow its first amount that Biden and Trump then replicated in under 8 years. Pierre Lassonde at the conference: global debt rose from $16T in 1980 to $314T in 2024; US federal debt from $1T to $37T over the same period.

The structural concern: the US is running a 6–7% GDP deficit at approximately 4% unemployment — what would historically be classified as full employment and a booming economy. In such conditions, public borrowing should be declining. Grant does not have a mechanism for when this becomes a crisis, and he is candid about it. The counter-argument: the dollar’s reserve currency status means no hard external limit currently; credit spreads are near all-time lows, meaning the market sees no immediate risk. Grant’s response: those conditions held “until this very moment” — but the amber lights (the 2019 repo market crisis, the March 2020 Treasury market dislocation) showed that even the deepest market in the world has its limits.

The political dimension: Grant expects the administration to attempt to subjugate the Fed after Jay Powell’s term expires in May 2026, installing a chairman who will implement the president’s preference for near-zero or negative rates. If successful: much lower money market rates, a weaker dollar, a much steeper yield curve (long-term rates rise even as short-term rates fall, as the market demands compensation for inflation and political risk on long-duration securities).

Röpke and inflation

Wilhelm Röpke (1899–1966) wrote in the 1950s about inflation as the national economy’s “sensitive organ” — damaged by consistent excess. The key passage: a “riot of claims and an insufficiency of goods produced to meet them.” People demand wages that exceed productivity growth, investment that exceeds savings, and imports that exceed exports; the government raises its claims on an already overstretched economy. The money supply expands to meet the claims that production cannot.

Röpke: “Just as there are organs in the human body in which, if consistently abused, ailments slowly but surely accumulate, eventually taking their revenge, so the national economy has its own equally sensitive organ. That organ is money. It becomes feeble and ceases to resist. And it is this infeeblement which we call inflation, a dilation of money, so to speak, a managerial disease of the national economy.”

Grant’s application: the era of paper money divorced inflation from its natural corrective. Under the gold standard, prices rose and then fell — gold outflows disciplined excess. Under a reserve-currency fiat regime, the mechanism is absent. “The dollar never regains the purchasing power it loses to inflation.” This explains why voters who lived through 2021–22 inflation did not accept that the problem was solved when the rate of inflation declined: they could see at the supermarket that prices did not return to 2019 levels.

Gold

Grant bought his first Krugerrand in January 1980 at approximately $850. He wore this purchase as “a badge of constructive humiliation” — it reminded him not to mistake narrative-building for certainty. He notes that the “permanent high plateau” argument of 2011 (after S&P downgraded US Treasuries from AAA to AA+) proved wrong; gold collapsed from $1,900 to $1,200 over the next few years, and gold mining shares fell up to 95%.

Gold is now near $4,000. Grant’s narrative for the current move: political pressure on the Fed (rates suppressed → dollar weakened → inflation risk) plus fiscal excess (unsustainable borrowing → eventual bond market protest). A prominent speculator told him at the conference: “Of course it’s a bubble.” Grant’s response: whether justified by fundamentals or not, bubbles can go further than expected. He holds gold, he acknowledges the narrative could be wrong, and he notes that anyone who “has a restless night’s sleep” from too large a position is carrying more than they should.

The Bitcoin contrast: Grant remains convinced the most efficient price for Bitcoin is zero. The “cryptobesotted Wall Street” — Vanguard considering crypto ETFs, $4T+ aggregate crypto market capitalisation, Bitcoin ETFs managing $142B — represents the kind of institutional adoption that historically marks late-cycle speculation. The administration’s promotion of crypto, including the president’s own “Trumpcoin,” Grant describes as “the scammiest thing” he has seen in his career.

Burke, Fox, and the 18th-century parliament

Grant’s book Friends Until the End centres on Edmund Burke and Charles James Fox — opposition parliamentarians for most of their careers, brilliant orators, and lifelong friends until the French Revolution broke them apart.

The key investor-relevant lesson: both were spectacularly bad with money (Fox amassed £140,000 of gambling debts in 1776 alone, equivalent to millions today; Burke’s estate was perpetually mortgaged) but this did not undermine the intellectual integrity of their public positions. Grant implies the contrast with modern finance: people who are disciplined and wealthy are not thereby wiser; people who are financially reckless are not thereby wrong.

The moral courage argument: Burke spent political capital defending the powerless (two men stoned in stocks for homosexuality; the destitute poet Crabbe who knocked on his door; the people of India against the East India Company) at enormous personal cost. He held unpopular positions — supporting the American colonists, opposing the French Revolution — and was shouted down, coughed at (“the dinner bell”), and ridiculed. His collected writings are his monument. Fox finally achieved his life’s greatest goal — outlawing the slave trade in Britain — on his deathbed.

The East India Company parallel: a corporation with its own army and navy, poorly paying its employees but allowing them to conduct personal side-businesses, producing predictable results: enrichment of the agents at the expense of shareholders and of the people they were supposed to serve. Grant draws this analogy explicitly with Buffett and Munger’s portraits visible on the wall — a nod to Munger’s dictum about incentives without naming it.

Jim Grant — speaker page
Decadent Finance — concept page
Value Investing — Grant’s conservative, defensive approach; connects to Marks’s fewer-losers framework
Risk Posture — Marks’s framework for calibrating aggressiveness; Grant embodies the permanent defensive posture
Compounding — the fiscal debt discussion; the AI capex race as anti-compounding (destroying capital at scale)
William Green — host; second Grant appearance