Observed Signal · Nov 5, 2026 · Technical Release · Source: SemiAnalysis · Impact: 4/5 · Sentiment: Positive
Google TPUv7 Challenges NVIDIA on AI Inference
SemiAnalysis released first official long-context, multi-turn agentic inference benchmarks (InferenceX/AgentX) showing Google's TPUv7 (Ironwood) delivers competitive or superior performance per dollar versus NVIDIA GPUs on several large models including GLM-5.2, Qwen3.5, Kimi K3, and Meta's Muse Spark Open. On GLM-5.2 FP8, TPUv7 delivers 1.2x-2.7x tokens per dollar of B200/B300/MI355X; on Qwen3.5 with llm-d disaggregated serving, TPUv7 beats GB300 FP8; on Kimi K3, TPUv7 delivers 1.15x-2.5x tokens per dollar of GB200/B200/B300/MI355X. NVIDIA's counter-submission on GB300 FP4 is still 1.64x worse per dollar than TPUv7 FP8 on Qwen3.5. The article highlights TPU's growing ecosystem adoption (Oracle, OpenAI, Anthropic, Meta) and contrasts NVIDIA's bureaucracy with AMD's agility in open-source collaboration. TPUv7 lacks native FP4 support until TPUv8 (Q1 2027), but competitive FP8 performance positions TPU as a serious alternative in AI inference.
TPU challenging NVIDIA in AI inference performance per dollar is a major shift in the AI hardware landscape, impacting cost structures for AI services and potentially reshaping the competitive dynamics in cloud AI infrastructure.
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 Polaris7 workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Google TPUv7 delivers up to 50% better performance per dollar than B200/B300 on Qwen 3.5 397B in FP8 (single-turn InferenceX result).
- On GLM-5.2 AgentX, TPUv7 FP8 delivers 1.2x to 2.7x tokens per dollar of B200, B300, and MI355X.
- On Kimi K3 AgentX, TPUv7 delivers 1.15x to 2.5x tokens per dollar of GB200, B200, B300, and MI355X.
- NVIDIA's GB300 FP4 counter submission is still 1.64x worse per dollar than TPUv7 FP8 on Qwen3.5 397B.
- Oracle, OpenAI, Anthropic, and Meta are customers of Google TPUs.
- TPUv8 expected Q1 2027 with native FP4 support.
Connected Companies & Entities
8 Entities mapped“Google's TPUv7 chip is performing well in agentic inference benchmarks.”
“NVIDIA's GB300 FP4 counter submission is still worse per dollar than TPUv7.”
“Meta is deploying TPUv7 in production using SGLang-TorchTPU and released Muse Spark Open.”
“OpenAI is a customer of Google TPUs.”
“Anthropic is a customer of Google TPUs.”
“Oracle is a customer of Google TPUs.”
“AMD has been more agile in open-source contributions and megakernel support.”
“SemiAnalysis published the benchmarks and analysis.”
Ontology 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 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.
Google launches separate TPUs for training and inference
Google Cloud announced its eighth-generation custom Tensor Processing Units (TPUs), splitting the family into two purpose-built chips: the TPU 8t for model training and the TPU 8i for inference. Google claims up to ~2.8–3x faster training versus prior generation Ironwood at comparable price, about 80% better performance per dollar on inference workloads, and the ability to cluster more than one million TPUs. Google said the TPUs will supplement — not immediately replace — Nvidia GPU offerings in its cloud, and that Nvidia’s Vera Rubin GPU will be available in Google Cloud later this year. Google also disclosed a collaboration with Nvidia to improve software-based networking (Falcon) for more efficient Nvidia system performance; Falcon was open sourced in 2023 under the Open Compute Project. The move positions Google’s cloud hardware as an alternative compute path for large AI workloads while maintaining interoperability with Nvidia-based stacks.
Receive all future signals for Google & competitors in your briefing
Never miss material M&A deals, leadership changes, account wins, or product rollouts. Polaris7 monitors 20,000+ primary sources 24/7, delivering verified telemetry directly into your executive briefing.