Goldman Sachs’ head of asset allocation research, Christian Mueller-Glissmann, delivered a sobering message on February 2, 2026: The massive wave of artificial intelligence (AI) capital expenditure (capex) that has fueled U.S. equity gains in recent years appears to be losing its ability to drive broader market performance. In an interview with Bloomberg Television, Mueller-Glissmann stated, “It feels like AI capex cannot really drive the equity market as much anymore… It’s not about the application and the adoption.”
This shift marks a pivot from the infrastructure-heavy phase of AI development—where hyperscalers like Microsoft, Amazon, Alphabet, Meta, and others poured billions into data centers, chips, and compute power—to one focused on real-world adoption, monetization, and measurable returns. Mueller-Glissmann highlighted that markets have increasingly traded AI in a “relatively evidence-based” manner, implying investors are demanding proof that these investments translate into sustainable profits rather than just hype-driven spending.
The comments align with Goldman’s broader 2026 outlook, which anticipates AI tailwinds evolving from capex dominance to broader adoption. Earlier analyses from the firm, including notes from strategist Ben Snider, underscore the profitability challenge: To maintain the high returns on capital that investors have grown accustomed to amid annual capex averaging around $500 billion from 2025–2027, AI-related companies would need to achieve an annual profit run-rate exceeding $1 trillion—more than double the consensus 2026 estimate of roughly $450 billion in income.
Consensus forecasts for hyperscaler capex in 2026 have climbed to $561 billion, a 38% increase from 2025 and up from $540 billion expected at the start of the earnings season. This escalation reflects ongoing buildouts, yet it raises questions about diminishing returns as spending scales without proportional revenue acceleration.
A recent example came from Microsoft, whose fiscal Q2 2026 earnings (reported late January) showed $37.5 billion in capital expenditures—a 66% year-over-year jump—yet Azure cloud growth of 39% (38% in constant currency) slightly disappointed some investors expecting closer to 40% or higher. Despite strong overall profits, shares dropped over 5% in after-hours trading, illustrating investor frustration with the decoupling between record spending and cloud revenue momentum.
This caution arrives amid broader market turbulence. Precious metals, particularly silver, remain volatile after a historic 30%+ plunge on January 30 (its worst single-day drop since 1980), triggered partly by a surging U.S. dollar following President Trump’s nomination of Kevin Warsh as Fed chair. Silver futures settled around $78–$83 per ounce in early February trading, extending losses from a speculative rally that pushed prices to record highs earlier in the year. Gold also weakened, though less dramatically.
Investor focus now intensifies on this week’s tech earnings: AMD reported Monday after the close, with Alphabet and Amazon due later. Any signs of slowing AI-driven growth or margin pressure from capex could amplify doubts. Meanwhile, FX strategists like NAB’s Ray Attrill note that despite Fed leadership uncertainty, markets are still pricing in about two rate cuts for 2026.
Goldman’s perspective frames a critical inflection point: While AI infrastructure spending remains robust and likely durable into 2026 (with some projections even higher, around $600 billion for top hyperscalers), the market’s next leg depends on whether adoption delivers the promised efficiencies, new revenue streams, and profitability. Without that “show me the money” moment, the era of capex alone lifting indices may indeed be waning, potentially leading to greater stock rotation, selectivity among tech winners and losers, and increased volatility as investors recalibrate expectations. For now, the AI narrative is maturing—from build-it-and-they-will-come to prove-it-and-they-will-stay.