Meta weighs a major AI chip order from Samsung, potentially challenging TSMC’s dominance

Meta Platforms is reportedly considering a large AI chip manufacturing order with Samsung Electronics’ chipmaking unit, Samsung Foundry, a move that could reshape the sourcing balance in the industry’s most expensive battleground: advanced processors used to train and run artificial intelligence models.

According to the report published by Yahoo Finance, Meta could place an order valued at as much as $6.5 billion, which the coverage describes as potentially its biggest AI chip order yet. The same reporting indicates Meta is weighing Samsung as an alternative to Taiwan Semiconductor Manufacturing Company, which is currently described in the story as the main or sole contract manufacturer for Meta’s chip needs.

The development matters because advanced AI chips are not only complex to design, they are equally difficult to manufacture at scale using leading-edge fabrication technology. For AI companies that build custom accelerators and then need them produced reliably in large volumes, the contract foundry is a key lever affecting supply, timing, and cost.

The Yahoo Finance article characterizes the possible Samsung switch as a significant procurement bet rather than a minor experiment, and it frames the negotiations as potentially large enough to represent a major share of Meta’s next phase of chip buying. Other web coverage in the same timeframe echoes the premise that Meta is in discussions with Samsung Foundry and that the dollar figure cited in the Yahoo report is central to the claim.

Samsung Foundry is the contract-manufacturing arm of Samsung Electronics, responsible for producing chips designed by external customers as well as internal chips for Samsung’s own product lines. If Meta expands beyond its current foundry setup, it would add a second large-scale production partner to a supply chain that has become tightly concentrated among a small number of top-tier manufacturers.

For Meta, the stakes are practical as well as strategic. Meta has been investing heavily in AI infrastructure, including the data-center compute required to train and operate large-scale AI systems. Those projects depend on steady access to high-performance chips and on the ability to ramp production fast enough to support model development cycles.

Still, none of the public reporting provides confirming details such as contract timing, chip specifications, production volumes, or whether the figure cited represents a signed agreement or a ceiling under negotiations. Meta has not publicly commented in the Yahoo report, and neither Samsung nor Meta has issued a primary disclosure about the alleged deal terms.

What to watch next is whether any of the parties involved provide confirmation, whether procurement documents or financial updates reference a new foundry relationship, and whether the deal, if finalized, corresponds to a broader shift in Meta’s AI chip strategy away from a single-source manufacturing approach. Even absent a signed contract, the direction of travel could offer clues about how quickly Meta wants to diversify supply for the next generation of AI capacity.