Flash

TopicJensen Huang Resigns as Nvidia CEO

Flash·Companies & Capital·2026-11-15 08:17

SemiAnalysis Names Nvidia's Seven Biggest GPU Customers; Six Build Their Own Chips

SemiAnalysis said on Nov 14 that seven companies — Google, Microsoft, Meta, Amazon, SpaceX, Anthropic and OpenAI — deploy the vast majority of Nvidia's GPUs, and all but SpaceX are building their own chips; its tests show Google's TPUs are on par with Nvidia's products.

A bright data-centre corridor with server racks and green status lights
AI-generated illustration

What happened

SemiAnalysis named the seven companies it sees deploying most of Nvidia's GPUs in a post on X on Nov 14: Google, Microsoft, Meta, Amazon, SpaceX, Anthropic and OpenAI, saying the seven account for the vast majority of GPUs being deployed. All of them except SpaceX are building their own chips.

The same post said that, from SemiAnalysis's benchmarking, the TPU is just as good as Nvidia, and that without Nvidia's backing of CoreWeave and other neoclouds, those industries would not exist and hyperscalers would have full control. In one SemiAnalysis test, Google's Ironwood TPU produced about 50% more output per dollar than Nvidia's B200.

Nvidia does not name its customers in its own filings. Its Q2 fiscal 2027 10-Q disclosed only that five direct customers held 22%, 14%, 13%, 11% and 10% of its accounts receivable, without identifying them. Nvidia's Q3 results are due Nov 17, three days after SemiAnalysis published its customer list; Q2 revenue was $96.2 billion, with $89.0 billion from the data centre business.

Key facts

  1. List: Google, Microsoft, Meta, Amazon, SpaceX, Anthropic and OpenAI account for the vast majority of Nvidia GPU deployments
  2. Own chips: All but SpaceX are building their own chips
  3. Benchmarks: SemiAnalysis said its tests show the TPU is on par with Nvidia; Ironwood TPU delivered about 50% more output per dollar than the B200
  4. Filings: Nvidia's 10-Q names no customers; Q2 revenue was $96.2 billion, with $89.0 billion from data centres
  5. Timing: SemiAnalysis posted on Nov 14; Nvidia reports Q3 results on Nov 17

Why it matters

the customer list puts Nvidia's demand concentration on the table, with its biggest customers also rivals building their own chips; the TPU cost figures landed just before the Nov 17 earnings call, putting customer and demand questions at the centre of the call.
Useful Tap if this story helped you

SourcesSemiAnalysis post, via FourWeekMBA (Nov 14, 2026); Kaimark (Nov 14, 2026). This article is compiled from public information and does not constitute investment advice.

From HIMIAO Horizon · himiao.ai

Grab todayOpenAI Codex for Students: $100 in Free CreditsHIMIAO Picks →

Follow HIMIAO Horizon

No sign-up · Unsubscribe any time

Email

Notifications

A ping on your phone or computer when the brief is out. No email needed.

HIMIAO PicksWeekly freebies: one email every Friday, a 5-minute read

Comments

  1. Loading comments…
Ask the cat