Meta is set to manufacture its latest AI-specific chips from September, aiming to cut costs amid global component shortages. The company has detailed plans for four new MTIA chips, which will power everything from ranking algorithms to broader AI tasks.

These modular designs are part of a larger push by Meta to secure substantial computing capacity for its AI projects. In April, it announced capital expenditures between $125 billion and $145 billion this year, with much going towards AI efforts. Other tech giants like OpenAI and Anthropic are also developing their own chips.

The shift reflects a broader trend in the tech industry where companies are investing heavily to maintain competitive advantages in AI. The deals Meta has secured with ARM, AMD, Amazon, Samsung and Broadcom illustrate its comprehensive approach to securing resources for its AI ambitions.

While this move is designed to save on GPU costs from Nvidia and AMD, it also signals a significant change in how tech companies might handle their AI workloads in the future. Expect more in-house chip development as the race intensifies.