
France’s 10-year government bond yield has climbed above 5%, exceeding the absolute levels recorded during the European sovereign debt crisis, as the gap over German 10-year yields reaches its widest since the euro’s introduction. In his report of 23 October, Goldman Sachs FICC trader Jonathon Chryssafis drew together this development and two other apparently separate market trends. Results from AI giants have revived doubts over whether the technology boom can last: capital expenditure has been reduced at Tesla, Meta’s AI expenditure has driven its credit default swap (CDS) spreads to record highs, and guidance from Lam Research has disappointed. Hedge funds, too, are withdrawing from technology stocks at an accelerating pace.

At approximately 40% below fair value against the dollar, the yen has the greatest valuation shortfall of any currency covered by Goldman’s GSDEER long-term fair value model. The bank sees a common consequence of these risks: reduced appetite for US dollar assets and the return of capital to Japan, supporting yen appreciation. Pricing in the options market remains subdued, however. Two-year at-the-money implied volatility for USD/JPY is just 9.9%, set against a 20-year average of 13.8% and an all-time low of 8.4%.
A sharp correction in AI valuations would have its centre of risk in credit rather than equity markets, particularly in bonds sold by lower-rated borrowers. Where that risk sits is central to the logic of Goldman’s report, although it is also one of its more readily overlooked conclusions.
Approximately $105 billion of US dollar-denominated project finance bonds had been issued in 2026 by 13 October, with almost all the proceeds earmarked for data centres and semiconductor or computing infrastructure. Compared with earlier records, this represents roughly four times the approximately $25 billion issued in 2020 before the AI boom, and nearly three times the record total for 2025. Goldman’s credit strategy team retains its underweight position in US dollar investment-grade technology bonds, alongside a neutral position in high-yield debt. It nonetheless makes clear that it expects performance among AI-related issuers to diverge further, most notably in project finance bonds.
Broader increases in AI infrastructure financing costs, should credit spreads widen, would begin to undermine the basis of the “US exceptionalism” narrative. That is the transmission channel Goldman identifies: weakening demand for US dollar assets sets its wider yen trade in motion.
A spread of 173 basis points between French and German 10-year yields forms another focus of the Goldman report’s discussion of the French bond market crisis. The gap is now greater than during the 2011 European sovereign debt crisis and wider than at any point since the euro was introduced. With France’s 10-year government bond yielding 5.08%, Italian yields sit approximately 50 basis points below their French counterparts. Such a reversal would have been barely conceivable in the 2011 crisis.
For spreads to narrow, Goldman’s strategists see the most straightforward catalyst as a fall in energy prices. Lower costs could shift the balance of risks for growth and fiscal deficits, help re-establish limited European Central Bank rate increases as the market’s baseline, and curb interest rate volatility. Investors could consequently be encouraged back into carry trades. Should energy prices fail to ease, the bank instead expects additional spread widening and increased pressure on the euro.
The risk emanating from France could therefore quicken the departure of European capital for safe havens other than the United States. Recovering domestic yields and Japan’s standing as the world’s largest net external creditor give it a structural advantage in attracting a share of those flows.
Concrete estimates of capital movements, rather than a loosely defined improvement in Japan’s economy, provide the basis for Goldman’s bullish assessment of the yen.
The potential impact of GPIF rebalancing is underappreciated. Japan’s Government Pension Investment Fund (GPIF) has approximately $2 trillion under management, Goldman observed in a September report. Were it to lift its domestic fixed-income allocation by five percentage points, roughly $100 billion would move into Japanese assets, accompanied by a matching volume of USD/JPY selling.
For context, that sum would account for nearly half the country’s annual current account surplus of approximately $210 billion, and would be substantially larger than the roughly $35 billion of net current and financial account flows recorded in 2025. GPIF could also establish a lead that other Japanese pension funds, insurers and long-term institutional investors might follow. Financial institutions and corporations could then begin unwinding yen-funded carry trades amassed over the past several years.
Further monetary tightening continues to find support in earnings figures. August brought a 1.5% year-on-year rise in Japan’s real cash earnings, the eighth successive monthly gain and the longest sequence in nearly a decade, while base wages increased 3.8%. If real wages keep rising, the argument for further Bank of Japan rate increases following its latest move becomes stronger.
The expected course of policy is embedded in market pricing, but not reflected in options valuations. By late October, overnight index swaps put the likelihood of a Bank of Japan rate increase that month at only roughly 6%–10%, against an 80%–89% likelihood of an increase by December.
September’s move by the Bank of Japan brought the policy rate to 1.25%. It was the second increase in three months, making this the quickest tightening cycle since 1990. The broad market view is for the rate to reach 1.50% at year-end or early next year, before rising again to 1.75% in March or April.
A rate increase in December thus belongs among the more likely outcomes, not the tail risks. Against that prospect, two-year implied volatility at 9.9% depicts a quiet market with annualised volatility under 10%. Goldman’s trade is built around the discrepancy between those readings.
There are fundamental grounds for the low level of implied volatility. A tight trading range near 158 has characterised USD/JPY recently, and realised volatility has run below its implied counterpart. Persistent selling of options has further squeezed the volatility premium.
A reinforcing cycle develops as a result. Sellers gain confidence from low realised volatility, and their activity depresses implied volatility further. This dynamic can hold only in the absence of new information reaching the market, however. In Goldman’s assessment, that condition is no longer being met.
The first change concerns the Bank of Japan: debate over further tightening increasingly centres on its timing rather than its likelihood. Markets appear to have adopted a December increase as their central case, given the probability of more than 80%. No commensurate rise in volatility has appeared in options prices, however.
A December Bank of Japan increase combined with no change from the Federal Reserve would reduce the US–Japan interest rate differential, directly diminishing the economic attraction of carry trades. Survey findings show that 94% of economists anticipate unchanged Fed rates in October, although every respondent expects a December hike.
The direction in which a surprise occurs is more significant still. A December move by the Bank of Japan could produce a sharp repricing of the yen if the market has not fully discounted it.
A second change is visible in options skew, even though aggregate volatility has not responded. Protection against a fall in USD/JPY continues to command a premium in risk reversals. Traders in options are thus paying to insure against the tail risk of yen gains while implied volatility overall stays low.
Taken together, pronounced skew and subdued overall volatility indicate recognition of the tail risk’s direction, but incomplete pricing of how large the move could be. The essence of Goldman’s recommendation is to capture that discrepancy by purchasing convexity at low implied volatility, instead of depending on correctly timed directional exposure.
Goldman’s revised forecasts provide the third element. The bank cut its three-, six- and 12-month USD/JPY projections on 25 September: the respective forecasts of 162, 163 and 165 became 158, 155 and 150.
Having been bearish on the yen in July, Goldman had turned bullish by September, reflecting a reappraisal of its macroeconomic framework. Its 23 October report gives practical expression to that change through a trade structure with a specified cost.
Taken as a whole, Goldman’s argument links three sources of upward pressure on the yen. Vulnerabilities in AI credit markets are eroding the basis of demand for US dollar assets, while French risks are hastening capital reallocation. Alongside these forces, structural repatriation to Japan is adding appreciation pressure, underpinned by GPIF rebalancing, the wage–price cycle and gradual Bank of Japan tightening.
Options markets nevertheless continue to charge an “insurance premium” for those combined risks that reflects the low-volatility equilibrium established during the recent spell of narrow trading ranges. At 9.9%, implied volatility still assumes an environment without incoming new information.