Chip Race Intensifies: DeepSeek Developing Own AI Chip, Apple Explores AI Chip Startup Acquisitions, Google Targets 6-10x Token Efficiency
Key Questions
Why is DeepSeek developing its own AI chip?
DeepSeek is building custom silicon to reduce Nvidia dependency, similar to Anthropic's Samsung partnership. This reflects a broader vertical integration trend among frontier AI labs.
What other companies are pursuing AI chip strategies?
Apple is exploring acquisitions of AI chip startups for inference and Private Cloud Compute. Nvidia offers a revenue-sharing compute model alongside $40B in AI equity investments.
What is the impact on the global chip race?
These moves intensify competition, add supply-chain complexity, and accelerate the shift away from sole reliance on Nvidia. Micron's $9.3B Japan plant expansion is part of the response.
China's DeepSeek is developing its own AI chip, mirroring Anthropic's partnership with Samsung for custom AI chip development. This signals a broader trend of AI labs verticalizing hardware to reduce Nvidia dependency. Challenges Nvidia's dominance and adds to the global chip race narrative, with implications for supply chain and competitive dynamics in applied AI. Nvidia offers revenue-sharing compute model with $40B in AI equity investments. Micron begins $9.3B Japan plant expansion. SambaNova raises $1B at $11B valuation. ZML releases free multi-chip inference server. New: Apple is exploring acquisitions of AI chip startups, signaling vertical integration for inference workloads and Private Cloud Compute. New: Google's Frozen v2 chip aims for 6-10x token efficiency over TPUs by embedding Gemini architecture in silicon, targeting 2028, though flexibility trade-off and execution risk remain.