Briev
Live
Business

Tech Titans Wrestle with Whether to Sell or Retain AI Compute Capacity

Meta, Google and Microsoft are debating how much AI-compute power to keep for internal projects versus selling to customers, as cash pressures mount.

Meta’s chief executive Mark Zuckerberg told investors that while most of the firm’s AI-compute will support internal model development, Meta intends to grow a commercial compute offering for large clients. The move comes as the company’s cash flow has plunged dramatically, prompting executives to consider monetising excess capacity. Google, meanwhile, raised its capex forecast and announced plans to acquire additional third-party compute and deploy its own tensor processing units to meet both internal AGI goals and external cloud demand.

Analysts warn that hoarding too much compute could drive enterprise customers toward rivals like Amazon or Microsoft. Microsoft’s finance chief Amy Hood echoed industry-wide shortages, noting that cloud demand continues to exceed supply, forcing a trade-off between immediate profit and future AI capabilities. The three firms thus face a strategic dilemma: sell compute now to shore up finances or retain it to stay ahead in the AI race.

Why it matters

How these firms allocate AI compute will shape cloud pricing, competition and the pace of AI development.

In this story

AI computecapexcloud businessTPUmonetizeAGIearnings callcompute shortage