Meta reportedly develops next-generation custom AI chip using TSMC's 5nm process node.

Meta Platform has announced its latest-generation Training and Inference Accelerator (MTIA) chip on April 10th, fabricated using TSMC's 5nm process. According to a report from Commercial Times, this move is expected to reduce Meta's reliance on NVIDIA's chips and enhance computational power for AI services.

The new chip, named Artemis, is an upgraded version of the MTIA introduced last year. Artemis offers triple the performance of the first-generation MTIA and is designed to support Meta's platforms like Facebook and Instagram with content ranking and recommendations. This development builds on Meta's broader AI strategy—last year, the company introduced AI models to compete with OpenAI's ChatGPT, reflecting its shift toward AI-driven services that require greater computational capabilities.

Meta's commitment to AI infrastructure is substantial. In October last year, Meta announced plans to invest USD 35 billion to establish infrastructure supporting AI, including data centers and hardware. CEO Mark Zuckerberg told investors, "In terms of investment priorities, AI will be our biggest investment area in 2024 for both engineering and compute resources." Meta's proprietary AI chips are deployed in data centers to power AI applications, with several ongoing projects aimed at expanding MTIA's application scope, including support for generative AI workloads.

The development of proprietary AI chips has become an industry-wide trend among major technology companies. Meta joins competitors like Amazon, Microsoft, and Google in internal AI chip development to reduce reliance on NVIDIA. Google recently unveiled its latest data center AI chip, TPU v5p, on the 9th, while Intel is targeting NVIDIA's H100 with its new AI chip, Gaudi 3.