Riskier Than Its Own Companies: What France's Bond Crisis Shares With the AI Capex Boom
French government bonds are now priced as riskier than more than a third of the country's own investment-grade corporate debt. The immediate cause is a French fiscal and political crisis, but the same global pool of savings governments borrow from is also funding an AI capex buildout now running near $800 billion a year.
Nearly €215 billion of French investment-grade corporate bonds, about 38% of that entire market, now trade at lower yields than French government debt of similar maturity. At the start of 2026 that figure was €12 billion. A government is supposed to be the safe asset a corporate borrower is priced against, not the other way around. Companies like L'Oreal and TotalEnergies have effectively swapped places with the French state in the eyes of the bond market. The immediate trigger is a French fiscal and political crisis. The more durable story is what else is competing for the same pool of global savings that sovereign borrowers depend on, and that pool is also now funding an AI infrastructure buildout running close to $800 billion a year.
When a Government's Own Bonds Trade Worse Than Its Companies'
The French 10-year yield came within reach of 5% in early October, its highest level since 2002, and the spread between French and German 10-year yields, the standard gauge of French sovereign risk within the eurozone, widened to roughly 150 to 160 basis points around October 1 and 2, the widest since the 2011 to 2012 eurozone debt crisis. Both measures have eased somewhat since, the spread back to around 140 basis points by the second week of October, but the move has not reversed. France is still paying a meaningfully higher premium than it was at the start of the year, and the market has not stopped treating a slice of French corporate debt as the safer bet.
A Fiscal Story That Didn't Start This Month
The Primary Deficit Nobody Fixed
The underlying number that distinguishes France from its peers is not the headline debt level, which at roughly 119% of GDP is lower than Italy's. It is the primary deficit, the budget gap before interest payments, which at 2.9% of GDP is the highest in the eurozone. Italy, by contrast, has run a primary surplus in most years since the early 1990s, about 0.8% in 2025. A government running a primary surplus is paying down its debt in real terms before interest even enters the picture. France has not managed that in any year since 2002. That track record, not the absolute size of the debt pile, is the reason markets have started extending Italy more benefit of the doubt than France, a role reversal that would have sounded improbable a decade ago.
A Government Report Even the Government Found Alarming
A July report commissioned by the French government itself, authored by a panel of independent economists, projected debt rising to roughly 130% of GDP by 2030 even if current deficit targets are met, and estimated that stabilising the debt would require on the order of €125 billion a year in savings, a figure some subsequent reporting has put closer to €140 billion. The report was the government's own commissioned assessment of its own trajectory. That it still produced a result markets found alarming is itself informative about how little room there is for a more optimistic read.
The Same Pool of Savings the AI Buildout Is Drawing From
France's crisis is specific to France. But it is unfolding inside a global bond market where the other large, fast-growing borrower is not a government at all. Goldman Sachs estimates the five largest hyperscalers will spend about $800 billion on capital expenditure in 2026, rising 54% to $1.2 trillion in 2027, above most Wall Street forecasts, before growth cools to about 12% and a $1.4 trillion total in 2028. More than a third of next year's capex may be debt-funded. Hyperscalers are, in the words of one Financial Times description, not price-sensitive borrowers: the spending plans get revised upward regardless of what it costs to finance them, which is a very different posture from a sovereign trying to convince a skeptical market it can hold a line.
That matters for France and every other sovereign borrower for a simple reason covered in more detail in an earlier piece on this page about the hyperscaler debt load itself: the capital hyperscalers are drawing from is not a separate pool from the one governments borrow in. Both are competing bids for the same global stock of savings, at a moment when one side of that competition is capped by voters' tolerance for austerity and the other is not meaningfully capped by anything except its own revenue, which Goldman estimates would need to reach roughly $300 billion a year just for the current spending to break even.
What This Means for a Portfolio That Treats Sovereign Risk as the Safe Part
The conventional portfolio assumption is that sovereign debt, especially from a large, developed eurozone economy, is the boring, reliably safe building block a portfolio is constructed around. France's own corporate bond market has just spent a year quietly repricing that assumption for a large slice of French credit, not because French companies got safer in any dramatic way, but because the government got relatively riskier. A fixed income allocation that treats "government bond" as a single, interchangeable category of safety is exposed to exactly this kind of repricing, and the same global capital competition that is straining French yields, a near-trillion-dollar-a-year private borrower with effectively no spending ceiling, is not going away once France's specific political crisis resolves. The pool of savings being split between the two is the same pool either way.
Whether French yields keep drifting lower from their early-October peak or widen again depends on French politics in the coming weeks. Whether sovereign debt in general keeps getting a smaller share of a savings pool now also feeding an $800 billion and rising AI buildout is a structural question that outlasts whichever government is in office in Paris.