Anthropic Is Building Its Own AI Chips
The company behind the Claude chatbot is hiring chip designers to make its AI faster and cheaper to run, following a path already taken by OpenAI, Google and Meta.
Key points
- Anthropic confirmed in August 2026 that it's forming a custom silicon team to design its own AI chips.
- The company is already in talks with Samsung as a potential manufacturing partner, per a report by The Information.
- Anthropic currently buys computing power from AWS, Nvidia and Google, but rising demand for Claude has made that harder to sustain.
- OpenAI unveiled its own chip, called Jalapeño, in June 2026, built with semiconductor company Broadcom.
- Google and Meta have both run on proprietary chips for years.
Anthropic wants to stop renting. The company is advertising roles on a "custom silicon team" that will design processors built specifically to run Claude, Business Insider first reported, with TechCrunch AI following.
Why would an AI company bother making its own chips?
Running a large AI model costs enormous money because companies must rent access to powerful chips from suppliers like Nvidia. Build your own chip and you can design it around exactly what your software needs, cutting waste and, eventually, cost.
Anthropic has deals with AWS, Nvidia and Google for computing hardware. As demand for Claude grows, buying time on someone else's chips puts a ceiling on expansion, and designing its own silicon removes that ceiling. The $10 billion cloud deal Anthropic signed with Volta on 4 August shows how aggressively the company has been shoring up supply even before its own chips exist.
Who else is doing this?
Several of the biggest names in AI have taken this road. OpenAI revealed its first in-house chip, Jalapeño, in June 2026, manufactured by Broadcom and built for inference workloads: the moment a user submits a query and the model generates a reply, the most frequent task an AI system handles.
Google builds its own chips internally, called TPUs (Tensor Processing Units), which power DeepMind's models. Meta has been developing its own MTIA accelerators for similar reasons.
A rough comparison of where the major players stand:
| Company | In-house chip | Status (mid-2026) |
|---|---|---|
| TPU (various generations) | Long in production | |
| Meta | MTIA accelerator | In development |
| OpenAI | Jalapeño (via Broadcom) | Unveiled June 2026 |
| Anthropic | Custom silicon (unnamed) | Hiring now |
What does this mean for people who use Claude?
In the short run, nothing. Chip design takes years before it reaches production. The benefits come later: faster responses, lower outage risk during busy periods, and potentially cheaper Claude-powered tools as Anthropic's costs fall.
Most AI companies that announce chip ambitions take three to five years to ship real silicon at scale. Hiring a team is step one of a long process, and Anthropic hasn't yet named a chip, a timeline or a confirmed manufacturing partner.
Anthropic didn't respond to a request for comment.
My read: The companies that control their own chips will carry a real cost advantage over those that keep renting, and that'll eventually show up in pricing. Watch how quickly Anthropic moves from job listings to actual silicon, because that gap is where most of these announcements quietly stall.
