Meta Platforms and Broadcom extended their strategic partnership to develop custom AI chips through 2029. The collaboration focuses on specialized processors for Meta's AI inference workloads. These chips power ranking systems and generative AI features across Meta's social platforms.
The agreement centers on the Meta Training and Inference Accelerator (MTIA) program. This partnership ensures continued development of future generations of Meta's custom silicon. The integration supports the massive scale of Meta's AI-driven services.
Meta projects capital expenditure on AI systems to reach between $115 billion and $135 billion in 2026. This investment secures a supply of optimized hardware to improve infrastructure performance and efficiency.