Meta extends partnership with Broadcom for AI accelerators

Meta is extending its partnership with Broadcom for the design of custom AI accelerators through 2029. The company also announced that Broadcom CEO, Hock Tan, will not seek reelection to Meta's board. Meta is set to initially deploy 1 gigawatt of its custom silicon Training and Interference Accelerators (MTIA), and will eventually deploy more gigawatts of chips based on Broadcom's technology. The move comes as tech companies scramble to make their own processing units to keep up with demand for AI data centers.

LinkedIn Talent Insights

Meta's hiring tells you why they need Broadcom. Meta added 11,992 people last year — but their three fastest-growing skills are all ML libraries (NumPy +25%, Pandas +24%, Keras +24%). Broadcom, by contrast, has 5,348 integrated-circuit designers and is still hiring chip-design specialists (Static Timing Analysis, Cadence, Logic Design all up). The 5-year extension is a talent-stack division of labor: Meta builds the models, Broadcom builds the silicon.

Top perspectives

David Berman · 2nd

Director, Product Management and Marketing, Enterprise Security

19h · 🌐

Building on their existing partnership, Broadcom will deliver technology supporting Meta Training and Inference Accelerator (MTIA) chips, with plans to extend through 2029. This technology will serve as the foundational backbone for Meta's deployment of state-of-the-art AI data centers.

Meta Partners with Broadcom on Custom AI Microchips

Priya Shankar · 3rd+

Semiconductor Industry Analyst

8h · 🌐

The MTIA roadmap through 2029 is the more important signal than the board reshuffle. Meta is committing to a multi-gigawatt AI infrastructure build — that's a TAM expansion for Broadcom's custom ASIC business...

Emily Okafor · 2nd

Cloud Infrastructure Engineer at Hyperscaler

5h · 🌐

Every hyperscaler now has its own custom silicon path: Google with TPU, Amazon with Trainium, Microsoft with Maia, and Meta with MTIA. NVIDIA's general-purpose moat is real but the AI capex is increasingly going to dual-source paths...