NextFin News - JPMorgan Chase Chairman and CEO Jamie Dimon delivered a split verdict on the two forces now reshaping global finance: the bond selloff that has pushed the UK's 30-year borrowing cost to its highest level since 1998 is a structural break investors should not mistake for a trading dip, while the artificial-intelligence buildout — roughly $700 billion of hyperscaler spending this year and on track to reach $1 trillion next year — is still in its early innings. The tension between those two calls, delivered in London on October 6 as UK bank stocks reeled from gilt-yield jitters and with a budget that could raise bank taxes weeks away, is the real story.
Speaking with Tom Mackenzie at JPMorgan's Tech Stars Conference in London, Dimon addressed the three issues defining the moment: how the AI boom gets financed, why the global bond market is selling off, and whether UK banks remain competitive. The setting mattered. Days earlier, on October 1, the yield on Britain's 30-year government bond crossed 6% for the first time since 1998 — the close reached 5.9352%, the highest since early 1998, while the 10-year climbed to 5.4033%, its highest since 2007, according to London Stock Exchange Group data. The FTSE 100 dropped 1.68% to 10,428.7, its biggest one-day decline since mid-May, and bank stocks fell 4.2% to their lowest level in more than three months. And in three weeks, Chancellor John Healey delivers his first budget, with a windfall tax on banks and oil companies still under consideration.
The Bond Selloff Is a Regime Change, Not a Trading Dip
The question Dimon's bond comments force is not whether yields can go higher — they already have — but whether the market is still thinking about the 2010s. For a decade and a half, the global financial system ran on a savings glut: aging populations, high Asian savings, and safe-haven demand kept long-term rates low even as debt climbed. Dimon's warning, repeated in interviews this year, is that the regime has flipped. "They could be much higher than they are today," he has said of interest rates. "We may have gone from a saving glut to not enough savings."
The mechanism is straightforward and it is already visible in the data. Government deficits are large in peacetime, quantitative tightening is withdrawing the biggest buyer of duration, and geopolitical risk plus elevated energy prices are demanding a higher term premium — the extra yield investors require to hold long-dated debt through uncertainty. The UK is the canary in that coal mine not because its fiscal position is uniquely bad, but because it is uniquely exposed: a large stock of long-duration gilt issuance, a pension and insurance sector built on the assumption of low yields, and a new government facing a fiscal hole it has pledged to fill without borrowing more.
That is why the 30-year gilt near 6% is more than a headline number. It is the discount rate for every long-lived asset in the British economy — mortgages, infrastructure projects, corporate debt, and bank balance sheets. A move from 3% to 6% on the long end does not just raise the government's interest bill; it halves the present value of cash flows twenty or thirty years out and forces a repricing across the entire capital structure.
The second-order effect is the part the market has not fully absorbed. Higher term premia do not revert the way a Fed funds rate does. A central bank sets the overnight rate; the market sets the term premium, and it does so based on the perceived supply of bonds, the credibility of fiscal policy, and the risk that inflation returns. Once investors decide that a country's debt trajectory is no longer self-correcting, the premium stays elevated until something structural changes — a credible fiscal consolidation, a growth inflection, or a central bank forced back into buying bonds. The 2022 gilt crisis, when the Bank of England had to step in to stabilize the market after the mini-budget, is the template for how quickly a term-premium shock becomes a liquidity event.
Is this cyclical or structural? The call here is structural, and the evidence is in the drivers rather than the price level. A cyclical selloff needs a short-term trigger — an inflation surprise, a positioning squeeze, a temporary supply spike — and it mean-reverts when the trigger fades. What has changed is the mix of savings and investment on a global scale: AI data centers are capital-intensive, defense spending is rising structurally, the energy transition requires upfront investment, and deglobalization is duplicating supply chains. Those are not cyclical; they do not self-correct. The post-1998 data supports the call: the UK 30-year yield has not traded near 6% in this century until now, and long rates have spent most of the period since 2021 above their 2010s average.
The falsifying signal is specific: if the UK 30-year gilt yield falls back below 4.5% and holds there for a full quarter without the Bank of England intervening as a buyer, the structural-break call is wrong and this was a positioning overshoot. That would require either a sharp growth disappointment forcing rate cuts or a fiscal consolidation credible enough to compress the term premium — neither is currently priced.
AI Spending: The Bill Is Real, and the Bubble Question Is Still Open
Dimon's AI comments sat in deliberate contrast to his bond warning. Spending across the hyperscaler ecosystem has more than doubled from about $300 billion last year to around $700 billion this year, and could reach $1 trillion next year, Dimon said at the JPMorgan India Conference in September. JPMorgan's own numbers are used to argue that the technology is already paying for itself inside the firm. The bank spends about $2 billion a year developing AI technology and measures about $2 billion a year in cost savings from it — a break-even Dimon has described as "the tip of the iceberg." Around 150,000 employees use the bank's internal large-language-model suite every week, embedded in risk, fraud detection, marketing, customer service, and idea generation.
The bullish case rests on capacity, not rhetoric. Data center vacancy is at a record low of about 1.6%, and roughly three-quarters of the capacity under construction is already pre-leased, according to JPMorgan's 2026 outlook research and CBRE market data. AI investment is currently around 1% of GDP — Dimon has equated the annual spending surge to roughly a 1 percentage point addition to GDP growth — and in prior general-purpose technology cycles such as electricity, railroads, and communications, investment peaked at 2% to 5% of GDP. On that measure, the buildout could still roughly double before it peaks. JPMorgan Asset Management estimates hyperscaler capital expenditure will reach about $697 billion in 2026.
But the financing structure is where the bond-market warning and the AI boom collide, and this is the second-order point most coverage misses. This buildout is not being funded out of retained earnings alone. It is being underwritten in a bond market that is simultaneously demanding a higher term premium, and the leverage is reaching levels that would have been unthinkable in a low-rate world. The record-setting $55 billion leveraged buyout of Electronic Arts, completed in August 2026, illustrates the point: JPMorgan committed $20 billion of debt, of which about $18 billion was expected to fund at closing, implying roughly 6x gross leverage at close, according to CreditSights analysis of the deal. The lending banks were reported to be preparing to sell down a substantial portion of that debt to outside investors — distribution risk sitting on top of a highly levered asset.
So is the AI boom cyclical or structural? The technology is structural; the financing of it may be cyclical. The adoption of AI inside enterprises — fraud models, customer service, code generation — is a permanent change in how work gets done, and JPMorgan's measured savings are evidence that the productivity channel is real rather than hypothetical. JPMorgan's outlook research notes that AI-related investment contributed more to U.S. GDP growth than consumer spending in 2025. But the capital expenditure cycle riding on top of that adoption is financed with debt whose cost has just been reset higher. If the term premium stays elevated, projects that cleared hurdle rates at 4% long rates do not clear them at 6%, and the capex cycle slows even if the technology keeps improving. The two claims are not contradictory: a structural technology shift can still produce a cyclical investment bust.
The falsifying signal for the "early innings" view is quantifiable: if data center vacancy rises above 5% while annual AI infrastructure spending stays above $800 billion, overcapacity has arrived and the cycle has turned. Vacancy above 5% with capex still climbing would mean supply is running ahead of the pre-leasing model that currently supports the investment case.
UK Banks: The Tax Fight Is About the Cost of Capital, Not the Tax Bill
Dimon's third topic — the state of UK banks — is where the bond call and the policy call meet. British banks already pay a 25% corporation tax rate, a 3% bank surcharge, a balance-sheet levy of between 0.05% and 0.1%, plus National Insurance, business rates, and sales taxes. The total tax rate on banks' UK operations was 46.4% in 2025, according to trade body UK Finance. Against that, London's tax burden is notably higher than Frankfurt, Dublin, and New York.
If you have a uncompetitive tax system, capital leaves your country.
Dimon said that in July, and he has tied the threat to a concrete asset: JPMorgan's planned £3 billion, 3 million-square-foot tower in Canary Wharf, which would house more than half of its roughly 23,000-person UK workforce. The bank's leadership met Prime Minister Andy Burnham and Chancellor Healey at Downing Street in September to argue against raising the tax burden in the October 28 budget.
The banks' argument is not that they cannot afford the tax. It is that the tax arrives at the worst possible moment in the rate cycle. A bank's equity value is a function of its return on equity relative to its cost of equity, and the cost of equity is anchored to long-term risk-free rates plus a risk premium. When the 30-year gilt moves from 3% to 6%, the hurdle rate for UK bank equity rises even before a single pound of new tax is levied. Add a windfall tax on top, and the UK becomes the place where the cost of capital is highest and the government takes the largest share of profits — the exact opposite of what a financial center competing for global capital wants.
The second-order consequence extends beyond bank stocks. If a tax pushes marginal capital out of London, the damage compounds through the listing market. A deeper, more liquid equity market is precisely what the UK needs to finance the AI and infrastructure buildout without relying on foreign lenders at 6% yields. Taxing the intermediaries that make listings work while trying to revive London as a listing destination is a contradiction the government has not resolved.
The counter-thesis is serious and it comes from the other side of the political divide. After years of bumper profits driven largely by higher net interest income, the argument runs, banks have benefited from the very rate cycle that is now squeezing everyone else, and a one-off windfall levy is a fair way to rebuild public finances without cutting spending. Trade unions and some lawmakers have made exactly this case. The strongest form of that argument is not that the tax is economically harmless — it is that the short-term efficiency cost is worth paying to fix a fiscal hole that, left unaddressed, is itself a driver of the gilt selloff Dimon warns about.
The answer to that counter-thesis is timing and design. A windfall tax announced into a bond-market stress episode signals that the government is choosing the politically easiest revenue rather than the economically least damaging one, and bond investors read that signal immediately. The falsifying signal for the banks' competitiveness warning is concrete: if the budget raises bank taxes and UK bank price-to-book multiples expand relative to US and European peers over the following twelve months, the capital-flight thesis is wrong. If instead UK banks trade at a widening discount and London's share of global listings keeps falling, the tax is doing exactly what Dimon says it will.
What to Watch: Three Horizons
In the short term, everything turns on the October 28 budget. The base case is that the government calibrates any bank levy to avoid a market shock — a targeted measure rather than a broad windfall tax — and gilt yields stabilize near current levels. The downside case is a broad-based levy combined with yields pushing through 6% on the 30-year, which would hit UK bank equities, housebuilders, and any domestic borrower with long-dated debt. The upside case is a reform package that pairs modest revenue measures with supply-side changes to listing rules and business taxes, allowing yields to roll back toward 5.5%.
Over the medium term, the beneficiaries of Dimon's worldview are the institutions that can price capital in a higher-rate world: US money-center banks with diversified funding, private credit managers that underwrite floating-rate debt, and the infrastructure and industrial companies that actually build data centers — the steel, cement, and power names Dimon has said you need to get to complete the AI buildout. The exposed are the UK-focused lenders, gilt-sensitive UK equities, and any highly levered borrower that refinanced on the assumption that the 2010s rate regime would return.
Over the long term, the structural call dominates: the savings glut is over, term premia are likely to stay positive, and AI adoption is a permanent shift in the production function even if the capex cycle overshoots. Those two facts together mean the next decade of finance looks less like the last one — cheaper money is not coming back, and the companies that win are the ones that earn their returns above a higher hurdle rate rather than the ones that borrowed cheapest.
The market has spent years treating every bond selloff as a buying opportunity and every technology boom as a bubble. Dimon's message is that both reflexes are now wrong at the same time: yields can stay higher because the world has run out of cheap savings, and AI spending can keep growing because the productivity payoff is already measurable. The investor who hedges one without understanding the other is hedged against the wrong risk.
更多独家洞察尽在 nextfin.ai.
