What is driving the simultaneous spike in sovereign yields across major economies, and does it signal the end of a dollar-anchored borrowing regime?

My position is that the yield spike should be read primarily as structural repricing, with cyclical inflation acting as the trigger rather than the root cause. A sovereign bond yield is the interest rate governments must offer to borrow; a term premium is the extra yield investors demand to hold longer maturities rather than roll short bills; AI capex is capital spending on data centers, chips, power and related infrastructure; reserve-currency status is the global willingness to hold and transact in a currency as a safe store of value and settlement asset; nominal rates are stated yields, while real rates subtract expected inflation. The core question is therefore not simply whether oil and CPI are high this quarter, but whether the global system is demanding a permanently higher real and term premium for duration, fiscal risk and dollar exposure.

The first strong evidence is synchronization across markets. Reuters reported that the latest selloff pushed borrowing costs to multi-decade highs, with the Middle East conflict raising energy-price risk while investors worried about inflation and ballooning government debt; Japan’s 10-year JGB yield rose above 3% for the first time in roughly 30 years. Separate reporting put the U.S. 10-year Treasury near 4.79% on September 1, while UK 30-year gilts reached about 5.88%, their highest since 1998. A one-country cyclical inflation scare would not normally produce simultaneous pressure in the U.S., UK, Germany and Japan; a common repricing of sovereign duration and fiscal credibility fits better.

Second, AI is no longer just an equity-market growth story. The Bank of England’s July 2026 Financial Stability Report states that hyperscaler year-to-date investment-grade debt issuance across currencies was broadly comparable in scale to UK gilt issuance over the same period, that Barclays projected $240 billion of 2026 AI hyperscaler investment needs financed through investment-grade credit, and that AI issuers accounted for 41% of non-refinancing U.S. high-yield issuance despite being only 1% of the high-yield index at end-2025. That is direct competition for balance-sheet capacity in credit markets, arriving when governments also need to issue heavily.

Third, the safe-asset foundation is weakening at the margin. The OECD says OECD debt-to-GDP is projected to rise to 85% in 2026 and that 10-year term premiums in major OECD countries had risen sharply since 2023, with the average term premium at the end of 2025 the highest in more than a decade. The IMF’s Fiscal Monitor adds that the U.S. Treasury safety premium has declined, raising the global risk-free benchmark. The Dutch central bank’s transfer of 86 tonnes of gold from New York and Ottawa to London, explicitly citing geopolitical unrest and crisis preparedness, is not proof of dollar collapse, but it is consistent with reserve managers diversifying operational and jurisdictional risk. Rogoff’s dollar-dominance thesis matters here: dollar dominance rests on trust, liquidity and U.S. institutional/fiscal capacity; erosion need not be sudden to raise term premiums now. What is at stake is the financing model of the last generation: if Treasuries no longer anchor global borrowing at unusually low real rates, governments, households and AI-heavy corporates all face a structurally higher cost of capital.

B’s strongest point is that the near-term trigger is real: the Iran-war oil shock and above-target inflation can plainly move front-end policy expectations and long yields, and AI capex is not yet a classic overlevered telecom-style bubble. That qualification matters because it prevents overstating a clean, monocausal “end of dollar” story. But it does not overturn my position, because the debate is about the fundamental repricing of the borrowing regime, not whether one week’s selloff had a cyclical catalyst. A structural repricing can arrive through cyclical shocks when investors discover that the fiscal and safe-asset backdrop no longer absorbs shocks at the old price.

The global-yield pattern is still hard to explain as only temporary inflation. Reuters reported on September 1 that the selloff hit major markets at once: Japan’s 10-year yield reached 3% for the first time since 1996, German yields reached 15-year highs, UK 10-year yields reached their highest since 2008, and the U.S. 10-year approached 4.8%, with the move reflecting both energy inflation and government-debt concerns. (live.euronext.com) If the argument were merely “oil up, central banks hike,” one would expect the risk-free anchor to hold better. Instead, the anchor itself is repricing.

On AI, B is right that low leverage reduces default risk, but default risk is not the central mechanism. The mechanism is capital-market absorption and duration supply. The Bank of England says hyperscaler year-to-date investment-grade issuance across currencies was broadly comparable to UK gilt issuance, Barclays projected $240 billion of 2026 AI hyperscaler investment needs financed through IG credit, and AI issuers represented 41% of non-refinancing U.S. high-yield issuance despite tiny index weight. (bankofengland.co.uk) Even if hyperscalers remain high quality, that much new private duration competes with sovereign issuance for the same long-horizon balance sheets.

Nor is dollar confidence irrelevant just because gold moved to London rather than Amsterdam or Beijing. DNB explicitly cited increasing geopolitical unrest, tradability, risk spreading, and crisis availability when moving 86 tonnes from North America to London. (dnb.nl) That is not dollar abandonment, but it is evidence reserve managers are paying attention to jurisdictional and operational risk. The IMF’s April 2026 Fiscal Monitor is more direct: it says the U.S. Treasury safety premium has decreased and that this effectively raises the global risk-free benchmark. (imf.org) Rogoff’s mechanism is the same: dollar dominance is weakened by fiscal unsustainability, weaponization of economic power, and threats to central-bank independence; if fiscal pressure triggers a crisis, market-share loss could accelerate rather than unfold slowly. (nybooks.com) B has shown there is a cyclical layer. The structural point is that each cyclical shock now clears at a higher required term premium because fiscal supply, AI credit demand, war risk, and safe-asset doubt are all pointing in the same direction.

B’s strongest case is that the immediate market mechanics look conventional: oil-war inflation raised expected policy rates, Warsh’s inflation focus fits the Fed’s 2% mandate, Japan and Germany each have domestic monetary stories, and the 2013 taper tantrum shows that a sharp term-premium episode need not imply dollar-regime erosion. B is also right that most AI capex is still supported by cash flow and equity, so the thesis should not be framed as an imminent hyperscaler solvency crisis or a simple one-for-one crowding out of Treasuries.

I still think A’s position is stronger because B explains the spark better than the fuel. A cyclical interpretation would be persuasive if yields rose mainly through expected short rates and then quickly mean-reverted as oil or policy expectations cooled. But the structural evidence is that the price of long-duration balance-sheet capacity has been repricing before and beyond the latest inflation shock: the OECD documents term premia rising since 2022 and reaching decade highs by end-2025, while the IMF states that increased Treasury supply is compressing the safety premium Treasuries traditionally commanded and thereby raising global borrowing costs. That is not merely “central banks may hike next meeting”; it is investors requiring more compensation to hold sovereign duration.

The AI point survives B’s leverage rebuttal. Low leverage means low default risk; it does not mean zero macro-financial impact. The Bank of England’s July 2026 report says hyperscaler year-to-date investment-grade issuance across currencies was broadly comparable to UK gilt issuance, cites Barclays’ $240 billion estimate of 2026 AI hyperscaler investment needs financed through IG credit, and notes AI issuers’ unusually large share of non-refinancing U.S. high-yield issuance. Even if the largest firms can self-fund much of the buildout, the external financing slice is large enough to absorb duration capital, bank balance-sheet intermediation, and private-credit capacity at the same time sovereign issuance is elevated.

Nor does the taper tantrum analogy defeat the structural claim. The Federal Reserve paper is useful precisely because 2013 was a monetary-communication shock in a world where Treasuries still enjoyed a large safety premium and post-crisis fiscal dominance fears were weaker. Today’s configuration is different: war spending and energy risk, persistent deficits, private AI infrastructure demand, and official reserve managers hedging geopolitical/jurisdictional exposure all coincide. DNB’s 86-tonne gold transfer to London is not dedollarization by itself, but DNB’s stated rationale—geopolitical unrest, tradability, crisis preparedness, and risk spreading—is consistent with marginal erosion in unconditional trust in reserve infrastructure. The dollar regime need not collapse for the borrowing regime to end; it only needs Treasuries to lose enough exceptional scarcity and trust premium that the world clears at structurally higher real yields.