Meta is set to begin production of its latest AI-specific chips in September, according to an internal memo cited by Reuters. The company is working to lower its GPU costs amid an unprecedented component shortage. At least one chip passed its testing phase in about six weeks, the memo noted. Meta is collaborating with Broadcom on the chip design but will manufacture them using Taiwan Semiconductor Manufacturing Company (TSMC). The company is also sourcing RAM from Samsung, storage from Sandisk, and fiber-optic equipment from Sumitomo Electric, according to the report.
Meta detailed four new chips developed under its Meta Training and Inference Accelerator (MTIA) program in March. Some of these chips are already in deployment or will be this year or next. The company is adopting a modular approach to chip design, anticipating evolving AI needs. The chips are expected to help Meta reduce its reliance on GPUs from Nvidia and AMD, although it still plans to purchase from those providers. The MTIA chips will be used for training models for ranking and recommendation algorithms, broader AI workloads, and inference for its applications.
Meta has been producing its own AI chips since 2023 and has invested heavily in securing compute capacity for its AI efforts. The company expects to spend between $125 billion and $145 billion this year, much of it on AI infrastructure. It has also signed deals with ARM, AMD, and Amazon for compute resources. Meta declined to comment on the matter.
Source: techcrunch
