Whalesbook
Planning for In-House AI Silicon
Anthropic is reportedly beginning to explore designing its own artificial intelligence chips. This is more than just a response to current hardware shortages. With Anthropic's annual revenue run-rate now surpassing $30 billion, up from about $9 billion at the end of 2025, the need for powerful and scalable computing has grown significantly. Developing its own chips aims to give the company more control over performance, supply chain stability, and costs, which are crucial for maintaining its leading position. Currently, Anthropic uses hardware from Google and Amazon, but designing custom silicon suggests a desire for greater independence.
AI Leaders Pursue Custom Hardware
This move aligns with a broader trend among major AI developers. Companies like Meta Platforms and OpenAI are also investing heavily in their own AI hardware. Meta is using its own MTIA chips, developed with Broadcom, for its large AI tasks. OpenAI has partnered with Broadcom on a major $10 billion project to design its first custom AI processors, with production expected in late 2026. These efforts show how important it is for AI companies to control their hardware to rely less on suppliers like Nvidia, which dominates the AI chip market.
Significant Risks in Custom Chip Development
Developing and manufacturing advanced AI chips is extremely expensive. Industry estimates show that designing 3nm chips alone can cost $400 million to $600 million. Beyond design, manufacturing costs for high-end chips like Nvidia's H100 can exceed $3,320 per unit, with complex superchips costing over $13,000. There's also a major risk that custom chips could become outdated quickly, failing to keep pace with innovations from established players. Meta's reported development struggles offer a cautionary example, as lengthy chip design cycles could create bottlenecks for a fast-growing company needing immediate scalability. Success also hinges on securing reliable manufacturing partners and navigating complex production processes. Additionally, Anthropic faces ongoing regulatory scrutiny in the U.S., which has been noted as a potential supply chain risk, potentially impacting revenue and creating uncertainty for business clients.
Strategic Goals for Growth
Anthropic's projected $30 billion revenue and growing base of large business clients show strong demand requiring continued investment in computing power. Exploring custom chip design, though in its early stages, could help Anthropic lower costs and improve its supply chain, similar to strategies used by major cloud providers. These early steps highlight Anthropic's ambition to manage key parts of its operations. How well the company handles the large costs and technical challenges of chip development will be key to its future success in the AI industry.