Hyperscalers commit $725B to AI infrastructure

Palantir and other enterprise software vendors are reporting accelerating AI adoption and capex commitment from hyperscalers and large enterprises. Palantir noted that when hyperscalers are committing $725 billion to AI infrastructure, the only remaining question is which companies are not going all-in on AI and how long they can afford to delay. This messaging resonates across the market, from semiconductor designers to power management companies to enterprise software platforms.

The capex commitment is structural rather than cyclical: Amazon reported negative free cash flowCash generated after maintenance capex; the actual money the business throws off. of $18 billion following $44 billion in property and equipment purchases in Q1, yet the stock rallied 45% over six months as long-term investors discount near-term cash burn for future infrastructure returns. Flex CEO Revathi Advaithi is leaving to lead Flex's $6.5 billion AI infrastructure spinoff, signaling confidence in the scale and durability of the opportunity.

Data center operators and power infrastructure vendors are benefiting. Fervo Energy, a geothermal energy developer backed by Bill Gates, raised its IPOInitial Public Offering - a company's first public sale of stock. target to $1.82 billion from $1.33 billion, positioning itself to capture the power demands of AI cluster buildout. Duke Energy applied for Department of Energy loans representing potentially billions in customer savings, signaling that utilities are racing to upgrade transmission and generation capacity for data centers.

The principal risk to this narrative is that capex productivity disappoints. If hyperscalers deploy $725B over three years but realize lower-than-expected returns, capex guidanceCompany-issued forecasts of future financial performance. will be slashed and semiconductor demand will crater. Additionally, geopolitical restrictions on chip exports to China could redirect some capex spending patterns. For now, the market is pricing in an extended capex cycle with limited downside.