Market research forecasts the data-center AI chip market will reach $860 billion by 2030, driven by inference demand and custom-silicon adoption.

The global market for artificial intelligence chips used in data centers is projected to reach $860 billion by 2030, up from $124 billion in 2024, according to a report by the Overseas Economic Research Institute at the Export-Import Bank of Korea. The market is expected to expand at an average annual rate of 38 percent over the period as investment in AI infrastructure continues to accelerate.

Nvidia dominates the sector with a 78.2 percent market share in 2025, followed distantly by Google at 4.7 percent and Advanced Micro Devices at 4.1 percent. Intel holds 3.7 percent and Huawei 2.6 percent. The market remains highly concentrated among a small number of U.S. and Chinese companies, though the institute noted that shifts in AI workloads could create openings for other chipmakers.

Graphics processing units, or GPUs, remain the dominant technology in data-center AI applications but are relatively expensive and consume large amounts of power. Neural processing units, or NPUs, designed specifically for AI workloads can offer lower costs and reduced power consumption—particularly important for inference, where demand is expected to grow significantly as AI services become more widely deployed. This shift from training toward inference workloads represents a strategic opportunity where Korean chip designers are seeking to expand.

South Korean AI chipmakers FuriosaAI and Rebellions are developing second-generation NPUs for data-center applications. FuriosaAI began mass production of its second-generation RNGD chip earlier this year, while Rebellions plans to launch its REBEL 100 chip in the second half. According to the institute, Korean companies have developed competitive chip-design capabilities but still face gaps of more than three years against leading global companies in areas including capital, commercialization experience, and industry ecosystems.

The institute emphasized that early deployment and customer references will be critical as the AI chip market develops over the next three to five years. It called for measures to reduce risks for companies adopting domestic AI chips, including tax incentives and subsidies, and recommended government support for Korean companies seeking overseas customers and supply-chain opportunities. The Middle East was identified as a potentially significant market, with regional countries expanding AI infrastructure while facing high GPU costs, power and land constraints, and uncertainty over access to advanced chips.