
The 200GB Question: Which Model Weights Actually Need to Sit on a GPU
DeepSeek V4.1 Flash needs 567GB of GPU memory rather than 763GB because 196 billion of its weights are built to run from system RAM instead.

DeepSeek V4.1 Flash needs 567GB of GPU memory rather than 763GB because 196 billion of its weights are built to run from system RAM instead.

Amazon is adding 2 million more Nvidia GPUs to AWS just five months after committing to 1 million, even as it scales its own Trainium and Graviton silicon.

The Series A extension doubles Starcloud to a $2.3B valuation, but Falcon 9 winds down in 2028 and Starship has yet to prove rapid reuse.

Cerebras says its new CS-4 rack, built from three WSE-3 Turbo wafers, delivers 750 PFLOPS and up to 30x GPU per-user token speed. Shipments start this quarter.

Nvidia will initially guarantee under $120B of a proposed Ohio OpenAI data center, down from $250B, backstopping only phase one after investor pushback.

SEC filings show Nvidia holding a $21B SpaceX stake and backing OpenAI's Ohio data center, deepening AI's circular financing web.

Volta, incorporated in January 2026 by ex-Brookfield executives, signed a roughly $10B six-year compute contract for a 133MW Norwegian site.