Tech Stocks Just Closed Out a Brutal Quarter With a Two-Day Sprint — As the World’s Central Bank Watchdog Warns It Could All Unwind Fast

Three days before quarter-end, the Nasdaq was sitting on its worst week in over a year. Two days later, it had clawed back nearly all of it in a pair of sessions that pushed the Dow above 52,000 for the first time and sent the chip-heavy SMH ETF up more than 3 percent in a single afternoon. The whiplash captures, in miniature, exactly the dynamic the world’s central bank for central banks spent its newest annual report warning about: a market that keeps reflating its AI trade faster than it can resolve the underlying questions about whether that trade is built on solid ground.

As the second quarter of 2026 closed, global markets staged a sharp recovery in large-cap technology and semiconductor shares, even as the Bank for International Settlements used its Annual Economic Report, published two days earlier, to caution that the scale of AI infrastructure spending now constitutes a genuine threat to financial stability, not merely a question of whether individual AI bets pay off.

What Happened

Technology stocks ended the final full week of the second quarter with sharp gains after a turbulent month. On Monday, June 29, the Dow Jones Industrial Average closed above 52,000 for the first time, climbing roughly 307 points, or 0.59 percent, while the S&P 500 rose 1.18 percent to 7,440.43 and the Nasdaq Composite surged 2.07 percent to 25,820.14. The Nasdaq-100, the index more heavily weighted toward megacap technology names, advanced by a similar margin on the session. Alphabet led the gains, jumping roughly 5 percent on its first day trading as a member of the Dow Jones Industrial Average, having replaced Verizon in the index. The VanEck Semiconductor ETF gained more than 3 percent, with chip-equipment names Astera Labs, KLA, and Applied Materials posting some of the day’s largest individual moves.

The rally capped a month that had whipsawed between sharp selloffs and equally sharp recoveries, driven alternately by worries that AI infrastructure capital expenditure was outrunning realistic demand, a brief but severe tech rout after a stronger-than-expected May jobs report, and escalating tension between the United States and Iran that briefly rattled energy markets and risk appetite together. Each selloff over the preceding weeks had been followed within days by a recovery of similar magnitude, a pattern that has itself become a talking point among market strategists watching how quickly capital keeps returning to the same handful of AI-exposed names.

The BIS Warning: A Different Kind of Caution

The rally’s backdrop is what makes it notable. Two days before the Monday surge, the Bank for International Settlements published its Annual Economic Report, in which the Basel-based institution, often described as the central bank for the world’s central banks, warned that the enormous scale of AI-related capital spending is accumulating financial vulnerabilities capable of amplifying any future shock and spreading it from markets into the broader economy. BIS general manager Pablo Hernández de Cos framed the message as one of urgency, urging policymakers to act before any eventual reversal makes the necessary adjustment more painful than it would be today.

The numbers behind the warning are large even by AI-era standards. The five largest hyperscalers are on track to commit more than $1 trillion to AI-related capital expenditure across 2025 and 2026 combined, spending that the BIS says is increasingly outpacing the companies’ own earnings and free cash flow, pushing some toward heavier reliance on debt issuance. The report also flagged a structural shift in how that spending is financed, away from internally generated cash flow and toward credit markets, with non-bank financial institutions, hedge funds, money market funds, and open-ended bond funds, taking on a larger role in absorbing the resulting risk than traditional bank balance sheets historically would.

The Mechanism: Why Asia’s Chip Hubs Are the Specific Pressure Point

Long-Dated Contracts Lock In Risk

The BIS report singled out supply bottlenecks in power generation, electricity grids, and memory chips as a distinct vulnerability layered on top of the broader financing concern, one with particular relevance for the AI buildout’s reliance on Asian semiconductor manufacturing. The mechanism runs through long-dated supply contracts: as hyperscalers compete for scarce capacity in advanced chips, grid connections, and power generation, the BIS warns they are increasingly forced into multi-year commitments to secure that capacity, agreements that lock in spending obligations regardless of whether AI demand ultimately materializes at the pace currently priced into markets. That overcommitment, the report argues, leaves the companies doing the spending more exposed to a downturn than a simple read of their balance sheets would suggest, since the true scale of forward-dated obligations does not always show up cleanly in standard financial disclosures.

A Concentration Risk That Cuts Both Ways

For manufacturing hubs across South Korea, Taiwan, Japan, Malaysia, and Singapore, where semiconductor production and assembly are concentrated, the exposure runs in both directions. These economies have benefited substantially from the AI infrastructure boom, with memory chip demand and data center buildouts contributing meaningfully to regional growth and corporate valuations. But that same concentration means a sudden slowdown in hyperscaler capital spending, the scenario the BIS explicitly flags as a risk, would transmit directly and quickly into the same regional economies currently posting some of the strongest AI-linked gains, with comparatively little buffer given how tightly the supply chain has wound itself around a small number of customers.

The Backstory: A Pattern the BIS Has Been Tracking for Months

This warning did not emerge from nowhere. The BIS flagged related concerns about AI financing structures in its March 2026 Quarterly Review, where researchers Inaki Aldasoro, Sebastian Doerr, and Daniel Rees documented that hyperscalers, including Amazon, Alphabet, Microsoft, Meta, and Oracle, had issued more than $100 billion in corporate bonds in 2025 alone, much of it carrying maturities beyond five years. That earlier report focused specifically on opaque, off-balance-sheet financing structures that the BIS said could obscure the true scale of hyperscaler leverage from investors evaluating these companies on conventional balance-sheet metrics.

The June 28 Annual Economic Report extends that earlier, narrower concern about disclosure into a broader statement about systemic risk, explicitly comparing the current AI investment cycle to historical episodes of infrastructure overbuilding, including the 1840s British railway mania and the dot-com boom of the late 1990s. Both of those earlier episodes, the report notes, involved genuinely transformative technology that nonetheless saw investment outrun realistic near-term returns, a parallel the BIS draws not to dismiss AI’s eventual productivity potential but to argue that real technological promise and sound financing structure are separate questions that markets have historically conflated.

Reactions

Market reaction to the BIS report itself was muted relative to its content, with equities continuing to rally through the two trading sessions immediately following its publication, a gap that Bloomberg’s coverage of the report noted explicitly, observing that the assessment landed on the eve of the European Central Bank’s annual Sintra symposium, where policymakers were expected to scrutinize the same stability risks closely. BIS chief Pablo Hernández de Cos told reporters the institution’s broader concern extended beyond AI alone to a “higher probability of second-round effects” across inflation and financial conditions, language that places the AI capex warning inside a wider set of pressure points the BIS says now demand policy attention. Reuters columnist Mike Dolan separately characterized the report as a notably blunt break from the institution’s typically circumspect tone, even as he observed that an enormous amount of capital in the current AI buildout remains, in his framing, still essentially a matter of faith rather than demonstrated return. The BIS itself was careful to stop short of recommending that central banks tighten monetary policy specifically to cool AI-related spending, with Hernández de Cos noting that prescribing a specific policy response would be unwise given the scale of uncertainty still surrounding how the AI investment cycle ultimately resolves.

The Dispute: Bubble Warning, or Routine Caution?

There is a real tension between how the BIS report has been read by different audiences. Some coverage has framed it as a stark bubble warning comparable to historical busts; the BIS’s own language is more conditional, explicitly stating that AI could still deliver significant productivity gains over the coming decade and that its concern is about financing structure and concentration risk rather than a judgment that the technology itself is overhyped. The institution’s own scenario modeling includes paths where AI-driven productivity gains materialize as promised. The disagreement, in other words, is less about whether AI investment carries real risk, both sides agree it does, than about how much weight to put on the BIS’s more dramatic historical comparisons versus its more measured statements about genuine upside potential.

What Happens Next

The most immediate test will be whether the rally that closed out the second quarter holds into the July earnings season, when investors get their next concrete look at whether hyperscaler capital expenditure guidance is moderating or accelerating further. The BIS’s warning also lands alongside ongoing Nvidia-linked deal activity across South Korea and other Asian markets, agreements that continue to deepen exactly the kind of regional concentration the report flags as a structural vulnerability, suggesting the tension between near-term deal momentum and longer-term systemic risk is unlikely to resolve cleanly in either direction before the next BIS reporting cycle.

Why It Matters

The juxtaposition of a record-setting tech rally with a pointed financial-stability warning from the institution that exists specifically to flag exactly this kind of systemic risk captures where the AI investment cycle stands heading into the second half of 2026: genuinely transformative technology, real and growing corporate spending, and a financing structure that an increasingly broad set of expert observers, not just skeptics, now consider under-scrutinized relative to its scale. Whether that gap closes through stronger disclosure and more measured spending, or through the kind of abrupt correction the BIS explicitly modeled as a historical parallel, is likely to be the defining financial-markets question of the second half of 2026.

Sources

Bank for International Settlements, Annual Economic Report, June 28, 2026. Market data via CNBC and Trading Economics for the June 29, 2026 trading session.