Observed Signal · Nov 28, 2025 · Partnership · Source: SemiAnalysis · Impact: 4/5 · Sentiment: Positive
Google Commercializes TPUv7, Challenging Nvidia
SemiAnalysis describes Google’s shift from internal TPU use to commercializing TPUv7 (Ironwood) hardware and renting systems via GCP, highlighting Anthropic’s confirmed 1 million TPU order split between direct purchases and GCP rentals. The report argues TPUv7 closes much of the performance gap with recent Nvidia GPUs while offering materially lower total cost of ownership (TCO) in many training workloads, and details the TPUv7 hardware, 3D-torus ICI scale-up network, OCS-based optical routing, software stack changes (native PyTorch support, Pallas kernel support), and ecosystem partners (Broadcom, Fluidstack, TeraWulf, Cipher Mining). SemiAnalysis frames this as a meaningful merchant-silicon challenge to Nvidia, with implications for datacenter power, neocloud hosting, supply chains, and open-source compiler/runtime adoption required to broaden TPU external adoption.
Major cloud provider (Google) commercializing advanced AI accelerators and signing large external orders (Anthropic 1M TPUs) can materially change AI compute economics, supply chains, and competitive dynamics with Nvidia—impacting model training costs and datacenter strategies across the AI ecosystem.
Track Google Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Google is commercializing TPUv7 (Ironwood) and selling complete TPU systems to external customers in addition to offering TPU capacity via GCP.
- Anthropic confirmed a deal for ~1,000,000 TPUs: ~400k TPUv7 Ironwoods to be purchased (Broadcom to sell finished racks) and ~600k TPUv7 units to be rented through GCP (SemiAnalysis estimates ~$42B of RPO).
- SemiAnalysis reports Gemini 3 and Anthropic’s Opus/Sonnet models were trained on TPUs, and claims TPUv7 offers substantially lower all-in TCO (Google-estimated ~44% lower per-chip TCO vs GB200 servers; external pricing still shows meaningful TCO advantages).
- Ecosystem participants named in the Anthropic setup: Broadcom (chip/system supplier), Fluidstack (on-site setup and cluster services), and datacenter infrastructure providers TeraWulf and Cipher Mining.
- Google is investing in TPU software improvements for external users (native PyTorch TPU backend, Pallas kernel support, vLLM/SGLang integration) but has not open-sourced XLA:TPU compiler, runtime, and multi-pod MegaScaler code.
Connected Companies & Entities
8 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Google TPUv7 Ironwood Delivers Up to 50% Better Performance per Dollar
SemiAnalysis, an independent analyst firm, published third-party inference benchmarks for Google's TPUv7 Ironwood, comparing it against NVIDIA's B200 and B300 GPUs. The results show Ironwood delivering up to 50% better performance per dollar in apples-to-apples comparisons, with a 19-34% lower cost per token at typical interactivity levels. Key factors include Google's co-designed hardware and software stack, the emerging TorchTPU external stack providing native PyTorch support, and optimizations across kernels and serving engines (vLLM, SGLang). Google is externalizing its TPU infrastructure, selling chips outright and on Google Cloud, with Anthropic as a major customer. The TorchTPU stack is in private beta, open-sourcing around mid-October, and future work includes speculative decoding, disaggregated serving, and KV-cache offloading, positioning TPUs as a strong competitor in the AI inference market.
Google sharpens TPU advantage in AI compute race
Alphabet’s homegrown tensor processing units (TPUs) are gaining prominence as a cost- and energy-efficient alternative to Nvidia GPUs, powering Google’s Gemini models and fueling Google Cloud’s enterprise growth. Google announced eighth-generation TPUs with distinct variants for training (TPU 8t) and inference (TPU 8i), claiming up to 3x faster training and 80% better performance-per-dollar, and has expanded commercialization—renting TPUs via cloud, selling hardware to customers, and launching a TPU cloud joint venture with Blackstone. Major AI labs and enterprises, including Anthropic and Meta, are adopting TPU capacity. Analysts and executives say TPU monetization and efficiency advantages could materially accelerate Google Cloud revenue and shift compute economics in the AI era.
Google's TPU Fleet Dominates AI Compute Growth
An analysis by Gennaro Cuofano highlights a major structural shift in AI infrastructure: Google’s TPU fleet expanded 11.5× over seven quarters, with quarterly additions accelerating. The report states Google's TPU power draw now exceeds Microsoft’s entire AI compute stack, and that by Q4 2025 Google added more compute in a single quarter than xAI had built in total. These data points indicate a widening infrastructure lead for Google that could cascade into lower per‑token costs, faster model iteration, and a larger operational moat for products and services that depend on large-scale inference and training capacity. The piece frames the compute growth as a pivotal industry trend with broad implications for competition, capability, and deployment timelines across AI-dependent sectors.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.