Tesla and Nvidia are reported to bankroll a $2.8 billion Samsung substrate buildout

Tesla and Nvidia will give Samsung Electro-Mechanics 3.7 trillion won, about $2.8 billion, to expand production of advanced substrates for AI chips, Reuters reported, citing Korea's Maeil Business Newspaper. According to the report, Nvidia would pay around 2.6 trillion won and Tesla about 1.1 trillion won.

The money is aimed at flip-chip ball grid array substrates, known as FC-BGA. They are far less famous than Nvidia's GPUs, but the chips cannot work without them. A substrate connects a large, complex processor to the circuit board underneath and carries its power and signals. As AI chips get bigger and draw more power, these substrates get harder to make, which turns them into another possible bottleneck in a supply chain that is already struggling with shortages of chips, electricity and data center space.

The investment builds on an October 21 announcement in which Samsung Electro-Mechanics and Nvidia unveiled a strategic alliance to expand South Korea's capacity for specialized substrates. The companies plan new manufacturing lines in Sejong and Busan, with Samsung supplying substrates for Nvidia's compute chips and the networking hardware that links data center processors.

Neither Nvidia nor Tesla had confirmed the exact amounts in the reports we reviewed, so the figures should be read as reported rather than official.

Why it matters for developers: when the biggest buyers start prepaying for obscure packaging parts, it signals that AI hardware supply will stay tight for years, not quarters. That shapes GPU prices, cloud capacity and how quickly new accelerators reach the instances you rent. It also shows how far down the supply chain the race for compute now reaches.

What's next: watch for official confirmation of the deal terms, timelines for the Sejong and Busan lines, and whether other chip buyers sign similar supply agreements.

Building with AI? See this week's hot AI repos and how to use each one.