Anthropic Decides to Build Its Own Chips, Because Apparently AI Companies Need More Hobbies
Anthropic is joining the custom chip party, because relying on Nvidia is apparently too mainstream. They're hiring chip designers now, so expect results in a few years, maybe.
Anthropic, the AI company that keeps trying to make Claude a thing, has revealed it's getting into the hardware game. Specifically, it's hiring a 'custom silicon team' to design chips for running its models. Because what's the point of a cutting-edge AI company if it can't also dabble in semiconductor engineering?
Yesterday, Business Insider spotted a job listing for a senior engineer with experience shipping semiconductor designs. You can now find listings for a silicon engineer and a technical program manager, silicon on Anthropic's job board. A spokesperson confirmed the plans to both Business Insider and TechCrunch, clarifying that Anthropic will take a 'multi-chip approach,' meaning it'll still use other companies' hardware alongside its own designs as it scales up.
This confirms rumors that have been floating around for a bit. The Information previously reported that Anthropic was considering working with Samsung as a hardware manufacturing partner. So, you know, just your typical AI company considering a partnership with a tech giant to make its own chips.
Anthropic isn't alone in this quest. OpenAI recently announced its own custom chip, Jalapeño, designed for large language model inference in data centers, developed with Broadcom. Google has been running its models on its own hardware for a while, Meta has designed and deployed its own chips, and Mistral is reportedly looking into doing the same.
Why the rush to custom silicon? First, the industry is heavily reliant on Nvidia for hardware, and that's a strategic vulnerability. Nvidia's leverage is a bit like having your entire business depend on a single, very expensive, very in-demand supplier. Second, designing chips specifically for your models can lead to better performance. If OpenAI can reap the rewards of vertical integration, you can bet Anthropic wants that too.
Anthropic says its teams will co-design new hardware and models side by side. It has co-designed hardware with partners before, but now it wants to bring more silicon expertise in-house. This could give its frontier models a competitive edge as developers explore running cheaper, smaller, or open-weight models on their own hardware.
However, since Anthropic is still hiring key team members, it'll be a while before any benefits materialize. So for now, you'll have to stick with Claude's occasional hallucinations and hope that custom chips make it smarter, not just faster at being wrong.
The Good Times
News in your inbox.
One sardonic roundup, delivered on your schedule. Free. Unsubscribe whenever your tolerance for wit runs out.
Already subscribed but we never reach your inbox? Check your spam folder and hit 'Not spam' (or 'Remove from spam') to bust us out of junk-mail purgatory. You'll be helping everyone else too.
Don't open any of our emails for a month and you'll be automatically removed from the mailing list.
Rewrite Article
Select parts to regenerate with a fresh AI pass. Translations will be updated automatically.
Generate AI Image
Creates a sardonic version of the article image using OpenAI.