Reuters first reported in April that Anthropic was considering producing its own AI chips to address supply shortages. Now, The Information confirms the company is actively discussing a collaboration with Samsung regarding this hardware. Anthropic has not yet specified the chip’s intended use, its role within a server, or its projected power levels. When asked for comment, the company stated that a diversified hardware stack including chips from Google, Amazon, and Nvidia remains essential to its compute strategy. It added it had nothing further to say about the potential Samsung partnership.
This move follows a recent announcement by OpenAI, which teamed up with Broadcom to launch its own custom inference processor called Jalapeño. OpenAI claims this new chip offers better performance-per-watt than competitor solutions. While Amazon and Google already provide custom-built TPUs as part of their cloud services, Anthropic’s exploration suggests a desire for greater independence from Nvidia, the current industry leader. Samsung is already embedded in the AI industry as a major partner of Nvidia, producing chips needed to train or run AI models. The two companies are working on an AI chip factory in South Korea and have previously discussed partnering with Google on chip-making efforts.
- Anthropic has not confirmed the chip’s specific application or power specifications.
- Samsung acts as a key Nvidia partner while simultaneously exploring independent chip designs.
- OpenAI’s Jalapeño processor aims to improve efficiency over existing market options.
