๐—”๐—ป๐˜๐—ต๐—ฟ๐—ผ๐—ฝ๐—ถ๐—ฐ ๐—˜๐˜…๐—ฝ๐—น๐—ผ๐—ฟ๐—ฒ๐˜€ ๐—”๐—œ ๐—–๐—ต๐—ถ๐—ฝ๐˜€ ๐—”๐˜€ ๐—ฅ๐—ฒ๐˜ƒ๐—ฒ๐—ป๐˜‚๐—ฒ ๐—›๐—ถ๐˜๐˜€ $๐Ÿฏ๐Ÿฌ ๐—•๐—ถ๐—น๐—น๐—ถ๐—ผ๐—ป

Anthropic wants to build its own AI chips.

The company is in early talks about custom silicon. This move follows massive growth. Anthropic now has an annual revenue run rate above $30 billion.

Why does this matter?

Large AI labs face massive compute demands. Relying on outside suppliers creates risks. If suppliers run out of stock, your growth stops. Custom chips help companies control their own supply chain.

Key facts about this shift:

โ€ข Anthropic is looking at custom chips to improve performance. โ€ข Custom hardware can lower costs per inference. โ€ข This puts Anthropic in line with Meta and OpenAI. โ€ข The company still uses Google TPUs and Amazon chips for now.

Building chips is expensive. It costs roughly $500 million to develop advanced AI silicon. This includes hiring talent and testing designs.

Anthropic is also making big moves with existing partners. They recently secured a massive deal for Google TPU capacity. They plan to invest $50 billion in US computing infrastructure.

The company is deciding between two paths. They can stay a hardware customer or become a hardware developer. At their current scale, both options make sense.

Source: https://dev.to/autonainews/anthropic-explores-building-ai-chips-as-revenue-hits-30-billion-2bch

Optional learning community: https://t.me/GyaanSetuAi