
Samsung Electronics (005930.KS) is establishing itself as a key player in the artificial intelligence (AI) chip market on the back of its foundry (contract chip manufacturing) business. Following Tesla, the company has been confirmed to be expanding its role as a production base for the application-specific integrated circuits (ASICs) of global big tech firms including Anthropic and Meta, a development expected to move up the timeline for a return to profitability. The industry expects Samsung Foundry's mid- to long-term order backlog to approach 50 trillion won.
According to industry sources Monday, Meta is pursuing the design and production of next-generation ASICs worth more than 10 trillion won with Samsung Foundry. Meta's proprietary AI accelerator, the "MTIA," was produced by TSMC through its first and second generations, but the third generation unveiled this year has designated Samsung Foundry as its partner. In particular, the MTIA third generation is set to apply Samsung Foundry's most advanced 2-nanometer (nm, one-billionth of a meter) process and be mass-produced on a scale of hundreds of thousands of units. A Samsung Electronics official said, "Nothing has been finalized yet."
Meta is currently reviewing entry into a "cloud service" business that leases AI computing infrastructure to outside companies. The MTIA is expected to be used as the core chip there. With a goal of building data centers totaling 5 gigawatts (GW) by 2030, the company cannot rely solely on external chips. In response, Meta plans to operate an ultra-fast development system that releases next-generation chips every six months, moving through the third generation to the fifth generation chip next year.
For this "chip development speed race," Meta is reported to have established a collaboration system with Samsung Electronics' System LSI Division, which specializes in design. Although Meta also operates its own AI chip design organization, it concluded that its in-house workforce alone would face limits in handling the ultra-fast six-month development schedule. Meta is working on joint development with Samsung System LSI from the early chip architecture design stage.
Anthropic, a U.S. AI firm, is also reported to be reviewing ASIC development using Samsung Foundry's 2-nanometer process. Analysts view this as part of an "AI infrastructure internalization" strategy to reduce dependence on external chips such as Nvidia's graphics processing units (GPUs) and Google's tensor processing units (TPUs).

Anthropic is pursuing the long-term construction of its own AI data center on a scale of about 1 GW, with the investment alone estimated to reach approximately $50 billion (about 77 trillion won). The industry expects about half of the total investment to go into AI chips, meaning investment in semiconductors such as ASICs, DRAM, and NAND flash alone is expected to reach approximately $25 billion (about 39 trillion won).
As Samsung Electronics possesses integrated semiconductor capabilities spanning memory, foundry, and packaging, it is cited as the biggest beneficiary of Anthropic's AI chip production. In May, Samsung Electronics participated in Anthropic's $65 billion (about 100 trillion won) Series H investment round, establishing a strategic partnership.
An industry official said, "Beginning with Tesla's AI chip order last year, Samsung Foundry's orders for AI server chips are entering full swing." The official added, "We are also pursuing an automotive foundry contract with BYD, China's largest electric vehicle maker, so the order backlog could grow to around 50 trillion won," and said, "An operating profit turnaround is possible as early as the fourth quarter of this year."






