Everything About AI
NextFin News - Global equity markets are holding up, earnings have been strong, and the artificial-intelligence trade still has credibility - but beneath that calm surface, global markets are accumulating "pockets of fragility" that investors can no longer ignore. That is the message from Alessia Berardi, head of global macroeconomics and emerging markets strategy at the Amundi Investment Institute, whose firm's September outlook warns that bond yields have climbed to levels not seen in decades, fiscal credibility is under scrutiny, and the AI build-out is now feeding risk back into the long end of the debt market through channels that did not exist two years ago.
The tension is stark, and it sharpened this week. The Federal Reserve raised its benchmark rate by 25 basis points on September 16 to a target range of 3.75%-4.00%, its first increase in more than three years, and signaled that more tightening is coming; new projections point to the policy rate reaching 4.00%-4.25% by year-end. Yet equities have advanced on the back of a strong earnings season, and Amundi's Global Investment Views for September, published as of 3 September 2026, frames the moment as "credibility under scrutiny" - a summer cocktail of earnings data, geopolitical headlines, and policy events that has investors increasingly questioning both policymakers and corporate balance sheets. The question is not whether growth has broken. It is whether the market is pricing a cyclical wobble or a structural regime shift.
This is not a call for an immediate downturn. Amundi maintains a mildly risk-on stance, arguing that the broader economic environment remains moderately constructive as long as credit and liquidity conditions stay benign. But the margin for error has narrowed: the 30-year US Treasury yield has reached its highest level in almost two decades, the US Treasury has resorted to buyback operations to ease long-end pressure, and a joint Treasury-Bank of Japan intervention in the yen has reinvigorated the debasement trade, pushing the dollar lower and lifting gold toward $4,350 an ounce. The fragility is real because the old hedges no longer work.
The Surface Calm and the Long-End Stress
On the surface, the bull market has reasons to feel intact. The S&P 500 hit a record high in early August, and the Q2 earnings season delivered results strong enough to keep the AI narrative alive. Amundi's September views note that "equities have held up, even advanced, on the back of a strong earnings season, which has so far supported the credibility of the AI trade." In a July interview, Berardi put it more directly: "We continue to remain risk-on [and] are constructive on risky assets. The economic outlook has been quite resilient... recession fears have not materialised."
But the bond market is telling a different story, and the divergence is the heart of the fragility call. The 30-year Treasury yield reached its highest level in almost two decades, touching 5.35% in mid-September and, after the Fed's September 16 decision, settling near 5.32%. That move prompted the US Treasury to announce a double buyback programme, and Amundi's report is explicit about what it implies:
"The scale of the Treasury's intervention would need to be substantial to ease pressure on long-end real rates."
In other words, the buyer of last resort for long-dated government debt is now a fiscal authority managing its own funding costs, not a central bank with an inflation target.
The mechanism matters. Long-end yields are not rising simply because the Fed is hawkish. Short-term policy rates are one thing, and the market's expectations for the terminal rate are another; the Fed's own projections now point to 4.00%-4.25% by year-end. What has moved is the term premium: the extra compensation investors demand for holding duration risk over decades. That premium is being repriced by three forces that do not self-correct quickly - uncertainty over major central banks' policy paths amid fears of sticky inflation, rising concerns about funding needs, and questions about fiscal discipline. Amundi's chart of the period shows the 30-year yield climbing toward 5.3% while the dollar weakens, the classic signature of a debasement trade rather than a pure growth scare.
This is the first pocket of fragility: a long-end market that has lost confidence in the policy mix, not just in the inflation print. And it is structural, not cyclical. A cyclical rise in yields would follow a cyclical acceleration in growth and inflation and would mean-revert as growth slows. What is happening now is different: the supply of government debt has changed regime, the central-bank balance-sheet era that absorbed duration is over, and investors are demanding a permanent risk premium for fiscal credibility. That does not revert on its own.
The AI Trade's New Feedback Loop
The second pocket of fragility is the one the market has not fully priced: the AI build-out is no longer funded the way it was two years ago, and that change ties tech valuations directly to the debt market.
Amundi's September report makes the link explicitly:
"Q2 earnings season has highlighted that the infrastructure build-out is accelerating, but increasingly through debt financing and off-balance sheet positions. This is adding to long-end supply pressure and creating a feedback loop between tech valuations and debt markets that did not exist two years ago. Going ahead, the scrutiny of company leverage will be paramount."
Unpack the mechanism. In the early phase of the AI cycle, hyperscalers funded data-center capital expenditure largely from cash flow and equity issuance - expensive capital, but capital that did not add to the supply of long-dated bonds. Now, with capex plans running into hundreds of billions of dollars, companies are turning to debt and to off-balance-sheet financing structures to fund the build-out. Every bond issued to build an AI data center adds to the long-end supply that is already under pressure from Treasury issuance. That pushes long yields higher. Higher long yields raise the discount rate applied to the very tech valuations that justified the capex in the first place. And if yields rise far enough, the economics of the AI investment case itself come under review - which is why Amundi says the AI theme is "becoming more selective."
This is a second-order effect that most bull-market narratives miss. The conventional wisdom is that AI earnings growth justifies any level of capex. The uncomfortable question is whether the funding of that capex is itself tightening the financial conditions that the valuation depends on. Two years ago, the equity market and the bond market could tell different stories. Now, through the leverage channel, they are coupled. That coupling is the fragility.
Berardi's own positioning reflects the selectivity shift. In the July interview she argued that investors should broaden AI exposure "upstream and downstream" across subsegments and geographies - power infrastructure, industrials, and software in Europe and India - rather than concentrating in the US mega-cap names. "South Korea and Taiwan are not really a diversification out of the US," she said, noting that both markets are dominated by technology sectors tied to US demand and the AI cycle. The implication is that the AI trade's concentration risk is global, not just American.
Geopolitics, the Debasement Trade, and the Narrow Fed Path
The third pocket of fragility is geopolitical, and it feeds both inflation and the debasement trade. Tensions in the Middle East persist, with disruptions in the Strait of Hormuz still unresolved and Red Sea risks increasing. Amundi stresses that maintaining alternative routes for oil supplies and bypassing the strait are "crucial for the economy's resilience and for mitigating inflationary pressures." It also flags the risk of rising food prices linked to climate change and the trend in gas prices, with reserves low in Europe and Germany in particular.
Oil is the transmission channel. Brent crude traded around $104 a barrel on September 17, up more than 14% over the prior month and roughly 54% above a year earlier. Amundi has ascribed a geopolitical premium to its oil outlook, expecting Brent to average $80 to $85 a barrel in the near term. That premium is a one-way inflation risk: it can spike higher on an escalation, but it does not fall cleanly until the underlying disruption is resolved. Berardi said in a July interview that "the geopolitical risk is not really diminishing. Gold continues to be an important hedge" - a view reflected in gold's climb toward $4,350 an ounce even as real yields rose, an unusual divergence that signals policy-hedging demand rather than simple inflation hedging.
The fourth pocket is the policy path itself. The Federal Reserve, Amundi writes, "is navigating a narrow path: inflation remains above target, but the labour market does not show signs of overheating." That narrow path is the definition of a stagflationary tightrope - raise rates and you risk breaking a labour market that is not overheating; hold or cut and you risk losing credibility on inflation that remains above target. Chair Kevin Warsh struck a hawkish tone at Jackson Hole in late August, and the September 16 decision - a unanimous 12-0 vote for a 25-basis-point increase, with projections pointing to at least one more hike in 2026 - confirms that the path is tightening, not pausing. The unusual feature of this cycle is that the Fed is hiking into an economy that does not show overheating, which is precisely the configuration that makes long-duration assets uncomfortable.
Meanwhile, the Bank of Japan is being pushed toward a regime change it would rather avoid. Amundi's report argues that the recent currency intervention by the US Treasury and the BoJ has "reinforced the case for a regime change at the BoJ," raising expectations for the terminal rate in this hiking cycle. With Japanese Prime Minister Takaichi unable to pursue an all-in reflationary mix of loose fiscal policy, loose monetary policy, a weak yen, and capital repatriation, adjustment is expected to come first through the yen and monetary policy. Amundi expects the BoJ to deliver four hikes before mid-2027, lifting the policy rate to a neutral 2%. The bank concludes its September 17-18 meeting with a 25-basis-point hike to 1.25% widely expected. The risk is that the adjustment is implemented in a way that creates "tensions in global fixed income markets" - a polite way of flagging a carry-trade unwind that could transmit volatility from Tokyo to the Treasury market.
Cyclical Calm on Top of a Structural Shift
So which is it - a cyclical fluctuation that will mean-revert, or a structural regime change that will not? The answer is both, and separating them is the key to not misreading the signal.
The cyclical leg is real and it is why Amundi stays mildly risk-on. Growth has held up. Earnings have surprised to the upside. Recession has not arrived. Liquidity conditions remain benign. Credit spreads have not blown out. These are the classic ingredients of a mid-cycle pause rather than an end-cycle break, and history suggests that when credit and liquidity stay benign, equity markets can digest a lot of bad bond-market news. The market's immediate reaction to the Fed's September 16 decision illustrates the point: the S&P 500 and Nasdaq were initially higher after the policy statement, with the 10-year yield actually easing about 4 basis points to 4.96% and the 30-year yield to 5.32%, before indexes gave back gains as the session ran its course. Berardi's July comment - "recession fears have not materialised" - captures this leg precisely. If the cycle is doing the talking, the fragility pockets are manageable, and the mildly risk-on stance is correct.
But the structural leg is doing the louder talking, and it is why the phrase "pockets of fragility" is the right framing rather than "buy the dip." Three structural changes are in place simultaneously: the fiscal regime has shifted toward persistent deficit financing that raises the term premium permanently; the central-bank regime has shifted from being the absorber of duration to being a source of policy uncertainty; and the AI capex regime has shifted from equity-funded to debt-funded, coupling tech valuations to the long end. None of these self-corrects. A cyclical claim would require evidence of mean reversion - three historical cycles where long yields spiked and reverted while fiscal deficits normalized. That evidence does not exist in the current setup; deficits are not normalizing, and the central-bank balance sheet is not expanding back to absorb supply.
The practical implication is that the old playbook - long duration as the hedge against an equity drawdown - is broken. In the pre-2020 regime, a stock-market selloff triggered a flight to quality that crushed yields and made bonds the perfect hedge. In the new regime, a selloff triggered by fiscal or credibility concerns can send yields and equities down together, exactly the correlation breakdown that defines a fragile market. That is why Amundi's conclusion is not "add duration" but "diversify": "as risks build beneath the surface, we believe diversification is now more important than ever."
The Counter-Thesis: Why the Fragility Call Could Be Wrong
The strongest case against this reading is straightforward: the market has been wrong about fiscal doom before, and long yields can stay elevated for years without breaking equities. History supports this. Treasury's daily record of the 30-year yield begins in February 1977, and from that first reading until late September 1998 the yield never closed below 5% - about 5,400 consecutive trading days, spanning almost 22 years. Over those 22 calendar years the S&P 500 returned about 15.8% a year with dividends reinvested, and 19 of the 22 years were positive. The lesson is that the level of long rates, by itself, decides almost nothing. If earnings growth remains strong and productivity gains from AI offset higher discount rates, equities can grind higher even as yields sit at levels that look alarming in isolation. Under this view, Amundi's fragility framing overstates a funding-supply problem that the private sector can absorb, and the mildly risk-on stance should be more aggressively risk-on.
There is force to this argument, and it is the view held by investors who see AI-driven productivity as the dominant force of the cycle. But it rests on one assumption that the fragility camp does not share: that the funding of the AI build-out remains orderly. The moment debt-funded capex meets a bond market that refuses to absorb supply at current yields, the feedback loop Amundi describes flips from theoretical to operative. The counter-thesis is right until the term premium says otherwise.
The falsifying signal is specific and observable: if the 10-year Treasury yield holds above 5.0% and the 30-year holds above 5.3% for two consecutive months while credit spreads remain below 100 basis points and equity indices make new highs, the fragility thesis is wrong - it would prove that the market can absorb elevated long-end supply without a transmission to risk assets, and the structural-shift call should be downgraded to a cyclical wobble. Conversely, if long yields hold at these levels and credit spreads widen above 150 basis points or the S&P 500 breaks its prior support on heavy volume, the fragility is confirming rather than fading.
What to Watch and Who Is Exposed
The forward look splits by time horizon. In the short term - sentiment and liquidity - the key events are the Federal Reserve's September decision, now delivered as a 25-basis-point hike to 3.75%-4.00% with a hawkish tilt, and the Bank of Japan's September 17-18 meeting, where a 25-basis-point hike to 1.25% is widely expected. A hawkish Fed combined with a dovish BoJ would strengthen the dollar and ease the debasement trade; the opposite mix would lift gold further and pressure the long end. Oil is the wild card: a Brent print above $110 on any escalation in the Strait of Hormuz would reflate inflation expectations and force the Fed's hand.
In the medium term - fundamentals - the watch items are Q3 earnings and, specifically, the financing disclosures around AI capex. If companies show a material shift toward debt and off-balance-sheet funding, the feedback-loop risk rises. If capex reverts toward cash-flow funding, the fragility eases. Corporate leverage scrutiny, which Amundi flags as "paramount," will be the lens through which the AI trade is re-rated.
In the long term - the structural leg - the signal is fiscal: the trajectory of US deficit financing and the Treasury's reliance on buyback operations to manage long-end yields. If buybacks become a permanent feature of debt management rather than an emergency tool, the regime shift is confirmed. If deficits narrow and the term premium compresses back toward historical norms, the structural call is wrong.
Who benefits and who is exposed? Beneficiaries of the fragility environment are the assets Amundi and Berardi point to: gold as a hedge against policy uncertainty and geopolitical risk; selected emerging-market debt where carry remains attractive; shorter-duration fixed income in Europe; and the downstream AI beneficiaries in power, industrials, and software outside the US mega-cap concentration. The exposed are the opposite: long-duration government bonds in core markets, highly leveraged AI infrastructure plays funded in the bond market, and concentrated equity positions in Korea and Taiwan that offer no real diversification from the US cycle.
Amundi's bottom line is a stance, not a prophecy: stay mildly risk-on, but diversify as if the surface calm is temporary. The market's job now is to decide whether the fragility is a warning light or the new normal.
The fragility is not in the price level - it is in the coupling. For two years, equities and bonds could tell different stories. Now the debt funding of the AI build-out, the fiscal funding of the deficit, and the credibility funding of central-bank policy have tied those stories together, and a break in one transmits to all. That is why diversification, not duration, is the only hedge that still works.
Explore more exclusive insights at nextfin.ai.
