Resolution criteria

This market will resolve to YES if, by December 31, 2027, at 11:59 PM UTC, Anthropic officially releases or deploys an in-house, custom-designed silicon chip (developed by Anthropic's internal silicon/hardware team) to train or run its Claude AI models.

"Release or deploy" is defined as any of the following:

• Anthropic publicly announces that it is actively using its proprietary, custom-designed chips for training or running Claude in production.

• Anthropic or its manufacturing partner officially announces the successful tape-out, shipping, or deployment of the custom silicon designed by Anthropic's in-house team.

Exceptions & Clarifications:

• Standard off-the-shelf hardware from NVIDIA, AMD, or other providers does not count.

• Standard cloud-provider chips (such as AWS Trainium/Inferentia or Google TPUs) do not count, unless the chip is explicitly co-designed and customized by Anthropic’s own hardware/silicon division.

• If Anthropic co-designs custom silicon with a partner (such as Samsung, Broadcom, or Google) specifically as an Anthropic-designed proprietary chip, this will qualify.

The primary sources of truth will be official press releases or statements from Anthropic, or reports from credible tech and financial news outlets (e.g., Business Insider, Reuters, TechCrunch, or Bloomberg). If no official release or deployment is confirmed by the deadline, the market will resolve to NO.

Background

In August 2026, Anthropic officially confirmed it was building an in-house silicon team to design custom semiconductors for its Claude AI models to run them more efficiently and address scaling bottlenecks.