Meta is reevaluating its internal AI chip development timeline after prototype performance and cost challenges, potentially delaying next-generation designs. Simultaneously, Meta is negotiating a $6.5 billion supply deal with Samsung for external AI accelerators to bolster data center compute capacity.
Meta is reviewing its internal AI chip development timeline after performance and cost challenges emerged with prototype designs, prompting engineering teams to postpone next-generation launches and focus on optimizing existing architectures.
Meta is in talks to finalize a $6.5 billion agreement with Samsung to secure external AI accelerators manufactured by Samsung’s leading foundry, aiming to bolster data center compute capacity and mitigate in-house delays.
This dual strategy of combining bespoke R&D with large-scale external sourcing diversifies Meta’s AI hardware supply, ensures uninterrupted processor access for its expanding AI workloads, and helps manage costs as demand for compute-intensive models accelerates.
