NextFin

Deutsche Bank Says the Dollar Has Become an 'AI Dollar' as Haven Status Fades

Summarized by NextFin AI
  • Deutsche Bank argues the US dollar has become an "AI dollar", tied to America's AI-heavy equity market rather than flight-to-safety flows, meaning it may fall alongside stocks during tech-driven stress.
  • The dollar index fell 9.4% in 2025, its worst annual performance in decades, trading around 101.5 after touching a 52-week high near 103 and a low near 95.5, showing de-correlation from the S&P.
  • Amazon, Microsoft, Meta, and Alphabet signaled up to $700 billion in AI capex this year, while the S&P 500 Software and Services Index fell nearly 26% earlier this year amid AI disruption fears.
  • Citi and PIMCO disagree with the "AI dollar" framing, arguing AI supports USD resilience and the correlation shift reflects positioning unwinds rather than a structural regime change.

NextFin News - The US dollar's reputation as the world's safest harbor is being questioned by one of Wall Street's largest currency desks, and the reason is artificial intelligence. Deutsche Bank AG argues that the greenback has effectively become an "AI dollar" — a currency whose fortunes are now tied to the performance of America's AI-heavy equity market rather than to flight-to-safety flows — meaning investors can no longer assume it will rise when stocks fall.

The shift matters because it upends one of the most durable rules in global macro investing. For decades, the dollar has been the asset investors buy when fear hits. Deutsche Bank's FX strategy team, led by George Saravelos, says that relationship has broken down: when the source of market stress is US technology itself, the dollar falls alongside equities rather than rallying.

The Thesis: A Dollar Priced by AI, Not by Fear

"The risks to the dollar are a lot more two-sided than the market has been pricing over the last few weeks," Saravelos, Deutsche Bank's global head of FX strategy, said in a client note on Friday. The bank's argument rests on a simple observation: recent episodes have shown tech stocks and the dollar weakening at the same time — behavior inconsistent with a traditional haven currency.

The mechanism runs through concentration. US equity indices have become dominated by a handful of AI-exposed megacap technology companies. When negative news about AI arrives — whether it is questions about whether hundreds of billions of dollars in capital spending will ever earn a return, or a new AI tool that threatens the business models of incumbent software vendors — the selling is concentrated in US assets. Foreign investors who own those assets then face losses denominated in dollars, and rather than buying dollars as a hedge, they reduce dollar exposure altogether.

"When the source of negative equity news is in the U.S., and the rest of the world is doing better, it is entirely possible for the dollar to fall as equities are going down, just like the 2002 dotcom period," Saravelos said. "The less attractive the dollar as a portfolio hedge, the more incentive there is to reduce dollar exposure."

The 2002 comparison is deliberate and specific. In the aftermath of the dotcom bubble, US equities collapsed while the dollar also weakened — a rare episode in which the two fell together because the crisis originated inside the US technology sector while growth prospects elsewhere were comparatively better. Deutsche Bank is arguing that the AI buildout has recreated that dynamic: the boom is American, the capital spending is American, and therefore the bust risk is American too.

The Evidence: Correlation Has Broken Down

The bank's case is not that the dollar will fall in a straight line. It is that the statistical relationship investors rely on has dissolved. "It is often taken as fact that the dollar is a safe-haven: it rallies during risk-aversion," Saravelos said in an earlier note. "A simple chart of the dollar-equity relationship shows this not to be true. The average USD-equity correlation has historically been closer to zero, and over the last year, the dollar has once again de-correlated from the S&P."

That de-correlation is visible in the data. The dollar index has been volatile rather than steadily bid, trading around 101.5 in the current session, down about 0.65% on the day, after touching a 52-week high near 103 and a 52-week low near 95.5. The index fell 9.4% in 2025, its worst annual performance in decades, and has given back further ground this year. In the same session, the S&P 500 and technology-heavy indices also weakened — the simultaneous softness that Deutsche Bank flags as the tell. The technology sector, which carries the heaviest AI exposure, has underperformed the broad market as the AI capex debate has intensified.

The trigger for the latest round of stress is the AI capital cycle. Amazon, Microsoft, Meta, and Alphabet have collectively signaled AI capital expenditure of up to $700 billion this year, and markets have begun to demand evidence of returns. Software stocks have been hit hardest. The S&P 500 Software and Services Index fell nearly 26% earlier this year, including a 3.1% single-day drop in April, after Anthropic released AI tools capable of handling professional workflows that many enterprise software companies sell as their core product. Anthropic later restricted wide release of its most powerful model, Claude Mythos, after it uncovered thousands of cybersecurity vulnerabilities, deepening investor fears about the threat to traditional software firms. At one point earlier this year, more than $1 trillion in market value had been wiped from large technology companies before a partial recovery.

Beyond AI, the dollar has faced a separate structural headwind: the "sell America" trade that followed the imposition of global reciprocal tariffs in 2025, which drove foreign investors to sell off US assets including the dollar itself. That policy-driven channel is distinct from the AI channel, but both point in the same direction — a dollar that is less reliable as a defensive asset.

Why It Is Different This Time: Cyclical Wave Meets Structural Shift

The critical question for investors is whether this is a cyclical wobble in the dollar — the kind that reverses when sentiment improves — or a structural break that will not repair itself. The answer, on Deutsche Bank's reading, is that both forces are at work, and they need to be separated.

The cyclical leg is the AI capital-expenditure cycle itself. Spending on AI infrastructure is front-loaded and concentrated; if returns disappoint, the repricing is swift and US-centric. That is a short-to-medium-term driver that can ease once the spending overhang clears or once earnings begin to validate the investment. Cyclical calls of this kind require evidence of mean reversion — and here the counter-evidence is real. The dollar has survived multiple "end of dollar dominance" narratives over the past two decades, from the post-2002 weak-dollar trade through the 2008 financial crisis, when the greenback rallied precisely because there was nowhere else to go. In 2008, even a crisis that began on Wall Street sent investors into dollars, not out of them — the exact opposite of the pattern Deutsche Bank describes for an AI shock.

The structural leg is different, and it is the more consequential part of the thesis. It rests on three claims: first, that US equity concentration in AI is not a temporary distortion but a durable feature of the market; second, that foreign ownership of US assets is now large enough, and hedged enough, that a US-specific shock produces dollar selling rather than dollar buying; and third, that the global growth backdrop has improved enough to offer investors credible alternatives. On the third point, Deutsche Bank points to the Australian dollar, the Scandinavian currencies, and emerging-market FX as more attractive than the dollar in this environment.

That structural claim is where the debate really lives. If US AI leadership is permanent and productivity-enhancing, the dollar should strengthen on the resulting growth differential — which is the opposing view. If AI turns out to be a capital-intensive boom with uncertain payoffs, concentrated in overvalued US equities, the dollar becomes a risk asset rather than a hedge.

The Counter-Thesis: Citi and PIMCO Disagree

Not every major institution accepts the "AI dollar" framing. Citi argued in a December 2025 note that the dollar performed well precisely because of AI, not despite it. "Barring significant repricing in U.S. equities," the bank wrote, "we believe global markets are likely to remain in this regime, which basically means a more resilient USD." In that reading, AI is a US competitive advantage that supports the currency — the mirror image of Deutsche Bank's argument.

PIMCO has taken a middle position. Its strategists argue that the recent correlation pattern reflects positioning unwinds rather than a structural regime shift, amplified by US fiscal deficits and uncertainty around AI-driven productivity gains. In its 2026 global outlook, PIMCO maintained that the dollar is likely to remain the world's dominant currency for the foreseeable future. That distinction matters: a currency can lose some of its tactical hedging properties while remaining the global reserve asset.

The strongest version of the counter-thesis is straightforward: the dollar's haven status rests on the depth and liquidity of US Treasury markets, the rule of law, and the absence of a credible alternative — none of which AI capex has changed. In a genuine global panic — a war, a systemic banking event, a sovereign debt crisis outside the US — investors would still buy dollars and Treasuries. Deutsche Bank's argument applies specifically to an AI-driven, US-originated equity correction, not to every form of stress.

Second-Order Consequences: What the Market Has Not Priced

The first-order implication of the thesis is obvious: do not buy dollars as a hedge against a tech selloff. The second-order implication is more important and less discussed. If the dollar no longer reliably rises when US equities fall, then the classic diversification that relies on currency hedging becomes less effective, and global portfolios become more correlated in downturns. Losses on US equities would no longer be cushioned by gains on the currency hedge — they would compound.

A third-order effect runs through the Federal Reserve and US funding conditions. A weaker dollar during a US equity correction would be inflationary, because imports become more expensive in dollar terms. That would complicate the Fed's task: the very episode that hurts risk assets could also push inflation up, limiting the central bank's ability to cut rates in response. This is the opposite of the benign dynamic investors enjoyed when the dollar rose during stress, which imported disinflation and gave the Fed room to ease. For a market that has priced in a dovish Fed as the backstop for equity weakness, a dollar that falls into the weakness removes part of that backstop.

There is also a cross-asset channel. If the dollar becomes a risk-sensitive currency, assets that have benefited from a strong dollar — including emerging-market debt denominated in dollars and commodities priced in greenbacks — would see their correlations shift. Conversely, currencies of commodity exporters and economies with improving growth, which Deutsche Bank names, could become the new marginal hedge. Gold, which traditionally benefits from both dollar weakness and uncertainty, sits at the intersection of these forces.

What to Watch: The Signals That Would Prove It Wrong

Deutsche Bank's thesis is falsifiable, and investors should watch for the specific signals that would invalidate it. The first is the USD-equity correlation itself: if it reverts to a sustained negative relationship — the dollar rising reliably on equity selloff days for a quarter or more — the "AI dollar" framing is wrong. The second is the AI capex cycle: if hyperscaler spending begins to generate visible, high-margin revenue growth rather than merely building capacity, the growth-differential argument for the dollar reasserts itself. The third is the global growth backdrop: if growth outside the US deteriorates while the US holds up, the dollar's relative attractiveness returns regardless of AI.

The most direct falsifying test is a US-originated equity shock that is not AI-related. If the dollar rallies strongly during a non-AI US equity correction — a regional banking scare, a fiscal standoff, a credit event — then the haven bid is still intact and Deutsche Bank's thesis is confined to a narrow AI-specific channel rather than a regime change.

Outlook: Three Scenarios

Base case. The dollar trades in a range with a downward bias as long as AI capex concerns dominate and global growth improves. The currency is no longer a reliable hedge against US equity weakness, and investors reduce unhedged dollar exposure gradually rather than all at once. Beneficiaries include the Australian dollar, Scandinavian currencies, and select emerging-market FX. Exposed are unhedged foreign holders of concentrated US technology exposure.

Upside case for the dollar. AI spending translates into measurable productivity gains and earnings growth, US growth outperforms, and the dollar resumes a structural uptrend. This is the Citi scenario, and its trigger is sustained evidence of AI monetization in hyperscaler earnings.

Downside case for the dollar. An AI-driven US equity correction accelerates, foreign investors accelerate dollar hedging or outright reduction, and the dollar falls alongside equities in a 2002-style pattern. The trigger is a credible downgrade to AI return expectations from one or more hyperscalers.

Across time horizons, the picture splits. In the short term, sentiment and positioning dominate, and the dollar is vulnerable to AI headlines. Over the medium term, the path depends on whether AI capex produces earnings. Over the long term, the dollar's reserve-currency status — anchored in institutions and market depth, not in technology cycles — is unlikely to be overturned by any single theme.

The bottom line: the dollar has not ceased to be the world's reserve currency, but it may have ceased to be the hedge investors assumed it was. In an AI-driven correction, the safe haven could be the very asset that is falling.

Explore more exclusive insights at nextfin.ai.

Search, ask or analyze anything...