SemiAnalysis Founder Dylan Patel Warns of $200B Annual AI CapEx Burn by Amazon, Google
Dylan Patel, founder and CEO of SemiAnalysis, said Amazon and Google are preparing to spend $200 billion annually on AI infrastructure—money he characterized as intentionally "lit on fire"—in a new episode of the podcast series "In-Context Cooking."
Patel, who rose from anonymous Silicon Twitter chip analyst and rural Georgia beekeeper to become one of the most cited voices on AI infrastructure, also flagged risks around Taiwan supply chains, semiconductor export controls, and what he called "Nvidia's paranoia advantage" in the competitive chip market.
The analyst, who advises hyperscalers, hedge funds, and semiconductor companies while tracking metrics from GitHub commit share to fab bottlenecks, did not elaborate on the specific timeline or conditions under which the $200B spend would occur.
The episode marks the debut of "In-Context Cooking," a new show format where guests discuss industry topics while preparing meals.
![[AINews] Anthropic @ $19B ARR, Qwen team leaves, Gemini and GPT bump up fast models](/_next/image/?url=https%3A%2F%2Fwordpress-production-adfc.up.railway.app%2Fwp-content%2Fuploads%2F2026%2F03%2Fhero-35f607d8.jpg&w=3840&q=75)
Responses (0 )