Meta augmenting, not replacing, Nvidia/AMD with in-house AI chips

Meta Platforms (META) is developing its own artificial intelligence (AI) chips, but these are intended to augment, rather than replace, the specialized hardware it procures from external chipmakers like Nvidia (NVDA) and Advanced Micro Devices (AMD). This strategy indicates Meta's ongoing need for high-performance AI accelerators from third-party suppliers to power its extensive AI infrastructure and models.

This development matters because it clarifies Meta's capital expenditure strategy for AI. Instead of signaling a future reduction in orders from leading GPU manufacturers, it suggests that Meta's in-house chip development will complement, not substitute, its existing supply chain. This implies sustained or even growing demand for external AI chips as Meta expands its AI capabilities.

The mechanism at play involves Meta designing custom AI silicon to handle specific workloads or optimize certain aspects of its AI operations, potentially for efficiency or proprietary features. However, for the most demanding and general-purpose AI training and inference tasks, Meta will continue to rely on the advanced graphics processing units (GPUs) and accelerators provided by Nvidia and AMD.

This news primarily affects Nvidia (NVDA) and Advanced Micro Devices (AMD) by reinforcing expectations of continued demand for their high-end AI chips, suggesting a stable revenue stream from a major hyperscale customer. For Meta Platforms (META), it indicates a dual-track approach to AI infrastructure, balancing internal innovation with external procurement to support its AI model capital expenditures and data center buildout.