Google launches external sales of its AI chips, challenging Nvidia’s dominance

Executive summary: Google announced its first external sales of AI chips, securing orders from Anthropic and other firms worth billions of dollars. This marks Google’s entry into the AI hardware market as a direct competitor to Nvidia, potentially shifting supply dynamics and creating a new revenue stream for the tech giant.

Who is involved: Google, Anthropic and other AI firms as customers, Nvidia as the incumbent market leader in AI accelerators.

Likely next: Google will likely expand its chip offerings and seek additional cloud and enterprise customers, while Nvidia may respond with pricing adjustments or new product releases.

Google announced that it is selling its AI Tensor Processing Units to external customers for the first time, with orders from Anthropic and other firms amounting to billions of dollars. The move signals Google’s intent to become a direct competitor in the AI hardware market, a segment currently led by Nvidia’s GPUs. While the initiative could diversify Google’s revenue and accelerate AI deployment, the company still faces technical and ecosystem hurdles that have limited broader adoption of its chips.

What's next — scenarios

Base: Steady adoption (40%)

Google secures several external AI chip customers, generating modest revenue that diversifies its cloud business.

Google announces additional external customers by Q1 2027
Nvidia maintains average selling price for H100 GPUs
Google releases TPU performance benchmarks by mid‑2026

Upside: Rapid market share gain (35%)

Google’s chips achieve performance parity with Nvidia GPUs, attracting large cloud and enterprise deals that erode Nvidia’s lead.

Google’s TPU v5 exceeds Nvidia H100 in MLPerf training benchmark by end‑2026
A major cloud provider signs a multi‑year TPU supply contract
Nvidia announces price cuts on its AI GPUs

Downside: Limited uptake (25%)

Technical and ecosystem barriers restrict external sales, keeping Google’s chip revenue negligible in the near term.

Google delays external chip shipments beyond Q2 2027
Key AI firms report compatibility issues with Google’s software stack
Export control restrictions limit sales to certain regions

What to watch

Google Q4 2026 earnings release (expected mid‑January 2027) to disclose AI chip revenue
Nvidia GTC 2027 conference (March 10‑13, 2027) for new GPU announcements
Anthropic’s planned integration of Google chips into Claude model (mid‑2027)
U.S. Department of Commerce review of export controls on advanced AI chips (expected Q2 2027)
Google Cloud Next 2027 event (May 2027) showcasing TPU offerings

Timeline

2026-10-02 — Technologie: Google fordert Nvidia mit eigenen Chips heraus (Handelsblatt)

Analysis — what this means

Likely next events

Google to report Q4 2026 earnings on 2027-01-15, expected to disclose AI chip sales figures
Anthropic to begin pilot deployment of Google TPUs in its AI training cluster by 2027-06-01
Nvidia to unveil next‑generation GPU architecture at GTC 2027 on 2027-03-10
U.S. Bureau of Industry and Security to publish updated AI chip export control rules by 2027-04-30

Sectors affected

AI semiconductor market
cloud AI infrastructure
healthcare AI
AI model training

Historical parallels

Apple’s introduction of the M1 chip in 2020 shifted PC CPU market dynamics
Amazon’s Graviton2 processor launch in 2020 challenged Intel’s dominance in server CPUs
Microsoft’s development of Azure‑optimized FPGA offerings in 2021 increased competition in AI acceleration