Overview
- The Information reported that Meta is discussing renting Google TPUs as soon as 2026 and installing dedicated TPU hardware in its data centers by 2027, a potential shift from Nvidia GPUs that remains unconfirmed.
- Nvidia responded on X that it is “a generation ahead” and said its platform runs every AI model across cloud, on‑premise and edge environments, positioning GPUs as more versatile than custom ASICs like TPUs.
- Nvidia shares fell roughly 2.6% to 3.2% on the reports as Alphabet gained and approached a $4 trillion market value, reflecting a rotation toward Google’s chip ambitions.
- Google said it is seeing accelerating demand for both its TPUs and Nvidia GPUs, and Gemini 3’s TPU-only training has reinforced perceptions that TPUs can handle key AI workloads.
- The flare-up follows Nvidia’s blockbuster quarter and strong guidance and has renewed debate over AI valuations and financing, including Michael Burry’s critiques, which Nvidia addressed in a detailed memo to analysts.