TechCheck Google’s Nvidia Chip Alternative 4/7/26
Apr 7, 2026
Mackenzie Sigalos, a CNBC tech reporter who covers big tech, chips, and AI, breaks down Google’s TPU strategy and its cloud availability. She discusses TPU cost vs performance, how Google might capture market share, and the vertical integration with Gemini and cloud services. She also talks about pricing dynamics compared with NVIDIA and what that means for valuation.
AI Snips
Chapters
Transcript
Episode notes
TPUs Becoming A Core AI Revenue Engine
- Google’s TPUs have shifted from internal tools to a major revenue driver as customers like Anthropic and Meta lease or buy them as cheaper alternatives to NVIDIA GPUs.
- TPUs are less flexible but more efficient and lower cost for certain AI training and inference workloads, potentially enabling Google to capture ~20% of the AI market.
TPU Market Opportunity Could Rival Google Cloud
- Analysts estimate Google's chip business could become a roughly $900 billion opportunity, potentially outpacing Google Cloud in value.
- That projection rests on TPUs gaining share as firms seek cheaper alternatives to NVIDIA and leaning into Google's vertically integrated stack.
Anthropic's TPU Orders Illustrate Rising Demand
- Mackenzie notes Anthropic will more than triple its computing capacity from Google's chips by next year as a concrete example of rising demand.
- Broadcom's purchase orders tied to Anthropic show TPUs can undercut NVIDIA on price in real procurement data.