Particle.news
Download on the App Store

Hyperscalers Pivot as Google, SpaceX and Microsoft Race to Secure AI Compute

Record cloud backlogs, new long‑term compute leases and high‑profile researcher departures are tightening demand for power, memory and bandwidth, raising investor concern about heavy AI spending.

Overview

  • Alphabet said enterprise AI is now its primary growth driver, reporting Google Cloud revenue up 63% and a contracted backlog that nearly doubled to more than $460 billion.
  • SpaceX is turning IPO capital into multibillion‑dollar compute leases that give labs long‑term access to Nvidia GPUs and dedicated power capacity, positioning the company as a major third‑party lessor.
  • Microsoft’s CEO urged broader access to models and has rolled out lower‑cost options and tools that let customers pick among models to push down prices and reduce concentration of AI power.
  • Two senior DeepMind researchers recently left for OpenAI and Anthropic, intensifying a talent war that is shifting where frontier AI research and hiring happen.
  • Analysts now expect a $700 billion‑plus AI infrastructure capex cycle focused on three bottlenecks—power generation, memory chips and optical networking—which is benefiting suppliers such as Micron and GE Vernova and driving new financing and leasing strategies by cloud firms.