AI trading has become a "rubber band," top Goldman Sachs trader says: The question is "when will it snap."

```

Between the massive investment in AI infrastructure and the uncertainty of its returns, a rubber band is being stretched tighter and tighter.

The latest analysis by two core trading supervisors of Goldman Sachs EMEA equity business points out that hyperscalers such as Microsoft, Amazon, Alphabet, and Meta are betting on AI infrastructure at a pace that exceeds their operating cash flows. Whether this unprecedented wave of capital can generate enough returns has become the most crucial unresolved question in the market.

Mark Wilson, Head of EMEA Equity Hedge Fund Business, and Rich Privorotsky, Head of EMEA Equity Flow Intermediation, believe the core contradiction lies in: The speed of AI construction is historically rare, while the prospects for short-term monetization and returns remain highly uncertain.

Privorotsky summarizes the current situation: "The AI market has become a 'rubber band'—the question now is how much further it can be stretched." He also pointed out that signs of cracks in the AI economic model appeared for the first time last week: frontier models are spreading rapidly, and inference costs are dropping sharply. Meanwhile, hyperscalers’ debt markets remain under pressure. If one company takes the lead in cutting capital expenditure, it could trigger a chain reaction.

Unprecedented scale of capital expenditure, unclear path to returns

Wilson and Privorotsky note that hyperscalers are pouring hundreds of billions of dollars into AI infrastructure, often investing more than their operating cash flows, aiming to seize the strategic high ground for the next-generation computing platform—a logic similar to previous bets on cloud computing and search.

Optimists see this as buying key "options" for the explosive growth of future AI agents, enterprise tools, and new applications. As costs fall and adoption scales up, returns will ultimately be realized. Pessimists worry about delayed returns, potential supply gluts, increased pricing pressure, and if cheaper alternatives emerge globally, whether enterprise monetization can land in time faces a severe test—this also raises market concerns over valuation bubbles and concentrated position risks.

According to the Epoch Capability Index cited by Apollo Global Management economist Torsten Slok, which integrates various AI benchmarks, the gap between the capabilities of open-source models and closed-source frontier models has narrowed to about four months, and the catch-up momentum of other national models should not be underestimated.

Value chain shifts upward: "tollbooth" logic replaces hardware narrative

Privorotsky admits he originally expected falling token prices would compress returns across the AI industry chain—with two competitive API and model providers outside cloud infrastructure, triggering aggressive price wars—but the current pricing logic puzzles him.

He analyzed two possible market signals: If the market believes "cheaper intelligence will expand the addressable market, drive enterprise adoption to exponential growth and boost overall compute demand," hardware stocks should lead the rally; but actual performance is not so. He therefore infers that the market may be expressing another judgment: Value is shifting upstream in the industry chain, with platform companies that control client relationships, distribution channels, and workflows enjoying durable economic moats—rather than hardware vendors.

He also points out, current trends largely reflect position adjustments and sector rotation, rather than fundamental changes in the basics.

The S&P Index is "impressive", sector rotation bears all the burden

Privorotsky admits to being "impressed" by the resilience of the S&P 500: the broad market not only ignores geopolitical and energy tensions, but has remained strong even as semiconductor and hardware stocks have clearly corrected. He said if you’d asked him a month ago, "Can the index rise even if chip stocks fall?", his answer would have been "impossible".

The current market feature is: correlations are breaking down, and sector rotation has become the main force supporting the index. The S&P 500, after removing AI-related weights, has broken out, with funds shifting from hardware suppliers to 'tollbooth-type' companies driving AI adoption—which control demand, software, cloud infrastructure, and distribution channels, closely aligning with the long-term economics of technology.

Wilson and Privorotsky believe compute power will become commoditized as supply expands, and hardware stock valuations will likely contract before earnings forecasts are revised down. The demand-side controllers are the true destination of long-term economic value.

Second quarter earnings will be a key near-term litmus test

The two trading supervisors characterize the upcoming second quarter earnings season as a key near-term test of whether the "AI investment return path" can make substantial progress. Recent unwindings of momentum strategies have caused market pain, but they believe the shift in sector leadership over the longer cycle is the right direction.

Overall, the Goldman trading team characterizes this cycle as: high risk and high volatility are inevitable, but if hyperscalers can control the "operating system" position in the AI ecosystem, they have strong long-term potential. The tone remains cautious on valuation and timing, but maintains faith in the upside of technological transformation.

Risk Disclosure and DisclaimerThe market has risks; investment needs caution. This article does not constitute personal investment advice, nor does it take into account the individual investment objectives, financial situation, or needs of any user. Users should consider whether any opinion, perspective, or conclusion in this article fits their particular circumstances. Investing based on this, responsibility is at your own risk. ```