Samsung Electro-Mechanics to supply NVIDIA with AI chip substrates; Exynos 2700 output reportedly to rise 10%

Samsung Electro-Mechanics has announced a partnership with NVIDIA to supply ABF package substrates for AI processors and networking switches, with production from 2026. Separately, reports say Samsung Foundry will raise Exynos 2700 wafer starts by over 10% on its second-generation 2nm process.

Samsung Electro-Mechanics (SEM) has announced a strategic partnership with NVIDIA to provide high-end artificial intelligence (AI) processors and high-speed networking switches using advanced ABF package substrates. The South Korean firm has said it will roll out production starting in 2026 and increase output to address the rising demand for AI computing equipment.

Separately, South Korean outlets Chosun and TheElec report that Samsung Foundry is preparing to raise production of its next-generation Exynos 2700 chip on its second-generation 2nm process. Together, the two developments could strengthen the Samsung group’s position in AI and semiconductors.

Samsung to expand ABF substrate production for NVIDIA

Under the deal, SEM will supply ABF package substrates for NVIDIA’s AI processors and high-speed networking switches. These substrates connect the components inside a chip package and help high-end computing systems meet their performance demands.

SEM will invest in expanding capacity at its Sejong and Busan plants in South Korea to meet NVIDIA’s expected demand. The company expects supply to grow at a double-digit percentage rate every year from 2026. By 2028, production for NVIDIA is expected to account for a significant share of SEM’s total substrate capacity.

SEM CEO Chang Duck-hyun said the partnership will help deliver high-quality components for NVIDIA’s AI platforms. NVIDIA executive Debora Shoquist said the agreement would widen the company’s supplier base and increase production capacity.

The deal is expected to strengthen SEM’s position in the AI semiconductor market as demand for data centres and high-end computing grows.

Pragya Singha Roy