CNBC Television
Google's AI chip 'Ironwood' takes on Nvidia
11/6/2025, 5:23:36 PM
Economic Summary
- Cloud providers are increasingly competing on custom silicon — 'chip wars' — because owning hardware offers cost and performance advantages that drive cloud AI business (e.g., GOOGL's TPUs versus NVDA GPUs).
- Alphabet (GOOGL) is scaling TPUs as a strategic pillar: Google Cloud revenue grew ~34% in Q3 and the company plans roughly $93 billion in CapEx for the year to support AI infrastructure and wins like OpenAI, Meta, and Anthropic.
- Geopolitics and export controls matter: US vendors face challenges deploying advanced silicon broadly, while China is accelerating domestic AI chip efforts (notably Huawei), which reshapes global competitive dynamics.
- Other large cloud and chip players (AMZN with Trainium, MSFT with Maya, Broadcom/AVGO, AMD) are pursuing custom silicon, and recent AMD export-license wins suggest near-term revenue upside from China reopening.
Bullish
- GOOGL: Proprietary TPUs (Ironwood) give cost and performance advantages winning cloud AI contracts.
- AAPL: Potential Gemini-powered Siri upgrade could meaningfully improve Apple's AI consumer offering.
- AMD: China export licenses reopening may unlock new revenue streams for AMD.
Bearish
- NVDA: GPU dominance threatened by rising in-house silicon and rival chips.
- China's homegrown chips (e.g., Huawei) could reduce US vendor market share.
- Geopolitical export limits restrict deploying competitive silicon in key markets.
Bullish tickers
GOOGLAAPLAMD
Bearish tickers
NVDAHUAWEI
GOOGL
Bullish
Ironwood TPUs and Gemini integration provide cost/performance edges, helping win major cloud AI contracts and drive cloud growth.
Bearish
Significant CapEx ($93B) and required global scaling pose execution and geopolitical risks that could limit TPU deployment.
AMZN
Bullish
Has its own Trainium chips to compete in the custom-silicon cloud AI race.
Bearish
Amazon's Trainium may lag Google's TPU performance, risking lost AI cloud opportunities.
MSFT
Bullish
Investing in custom silicon (Maya) to support its cloud AI offerings.
Bearish
Microsoft's Maya silicon faces stiff competition from Google's TPUs and NVIDIA GPUs.
NVDA
Bullish
Still holds GPU dominance and maintains messaging asserting U.S. edge in semiconductor performance.
Bearish
NVIDIA's GPU leadership could be eroded by major cloud providers' in-house chips and faster cost/performance solutions.
AAPL
Bullish
Reportedly tapped Google's Gemini to power a Siri upgrade, which could meaningfully improve its AI assistant.
Bearish
Siri AI upgrades have been repeatedly delayed and partnership details with Google remain unconfirmed.
AVGO
Bullish
Actively trying to challenge NVIDIA and participate in the AI silicon market.
Bearish
Broadcom faces difficulty displacing entrenched GPU incumbents in AI infrastructure.
AMD
Bullish
CEO Lisa Su says recent export licenses should allow AMD to recognize revenue from China reopening.
Bearish
AMD still trails NVIDIA in AI GPU market share and must convert China reopening into sustainable revenue.
META
Bullish
Is a customer in Google Cloud's recent wins, indicating use of TPU-powered services.
Bearish
Dependency on third-party cloud silicon could create strategic vulnerabilities.
OPENAI
Bullish
Reportedly a cloud customer for Google's TPU infrastructure, benefiting from performance and cost advantages.
Bearish
Reliance on cloud partnerships could limit control over infrastructure costs or performance.
ANTHROPIC
Bullish
Listed among companies that have signed cloud contracts with Google for TPU access.
Bearish
May face competition for cloud partnerships and infrastructure control from large providers.
HUAWEI
Bullish
China is aggressively promoting domestic AI chips like Huawei's, enhancing local competitive strength.
Bearish
Chinese homegrown chips may face their own technical and export limitations outside China.
People mentioned
Mackenzie SigalosCarlSundar PichaiLisa SuJensen Wong