The AI rally is broadening beyond its biggest names, and its financing is getting harder to trace

What changed

The message from several large research houses this week is consistent, even if the emphasis differs. The AI theme is still described as a key growth engine, but the easy version of the trade, owning the same handful of names and assuming they all win, is no longer the whole story. UBS points to high expectations and differences in company outcomes as a source of single-stock concentration risk, and it argues that equity portfolios should be broadly diversified. Wellington makes a similar point in a different way, saying that AI exposure is not as simple as picking winners. m:…4371 m:…6336

The market action behind that advice is visible in the numbers UBS cites. The Philadelphia Semiconductor Index gained 2.8 percent on Monday while the Nasdaq rose 0.8 percent, a tech-led rise that still left room for a diversification argument. Bank of America adds that its equity derivative strategy team views the recent rotation from technology into defensive stocks as possibly having gone too far, based on its Bubble Risk Indicator. Wellington and Bank of America both describe the recent AI sell-off as largely a positioning-driven setback rather than evidence that the longer-term theme is breaking down. m:…4371 m:…6485 m:…6336

Where the money is moving

Power and resources are the clearest second leg in this account. UBS says structural demand for power and resources remains in place because of AI and electrification needs, and it points to longevity opportunities supported by aging populations and clinical innovation. Wellington says rising data center electricity demand in the United States is creating new opportunities for utilities. Bank of America puts a scale on the buildout, expecting total data center spending to reach 2.1 trillion dollars by 2028, more than double the forecast from last year, and it notes that NVIDIA efforts to integrate more of the data center stack have helped drive more than 40 billion dollars of mergers and acquisitions since 2024. m:…4371 m:…6336 m:…6485

Geography is the other diversification route. UBS says Eurozone equities are supported by manufacturing recovery and structural investment, and that Asian equity valuations are at multi-decade lows. Wellington says the technology sector in Asia trades at a discount versus global peers. The implication both houses draw is that the next leg of AI exposure may sit outside the most crowded names and markets, in utilities, in resources, and in regions where valuations have not already priced in the same outcome. m:…4371 m:…6336

The financing warning

A Banco de Espana occasional paper adds a caution that sits underneath the optimism. It says leading AI-related companies have increased co-investments with non-bank financial institutions, particularly through venture capital, and that non-bank involvement in joint ventures has been substantial, driven by data center expansion. The paper then warns that the resulting interconnectedness may amplify vulnerabilities, particularly amid rising leverage and the growing complexity of AI-related investment and financing structures. m:…0038

The most pointed part of that warning concerns transparency. The paper says circular investment and financing arrangements, cross-shareholdings and special purpose vehicles may reduce transparency regarding ultimate funding sources and risk concentrations. That matters for a banker because the headline deals look straightforward, while the funding chain behind them may not be. UBS cites three such headline transactions in a single note: Broadcom arranged a 63 billion dollar deal to finance the purchase of chips, SpaceX signed a 40 billion dollar equipment-finance lease transaction, and Meta increased its equity fundraising to 97.5 billion dollars. The question the paper raises is not whether the spending is real, but whether everyone in the chain can see who ultimately bears the risk. m:…0038 m:…4371

What to watch

UBS says hyperscalers September-quarter results will provide evidence on the sustainability and monetization of AI demand. That is the test that separates a positioning unwind from a fundamental break, in the framing shared by Wellington and Bank of America. If monetization keeps pace with spending, the broadening trade has a fundamental leg to stand on. If it does not, the concentration risk UBS flags becomes the main story again, and the opaque financing structures described by Banco de Espana become harder to ignore. For now, the research consensus in these notes is to spread exposure rather than to exit the theme, and to ask harder questions about how the buildout is funded. m:…4371 m:…6336 m:…6485 m:…0038