Tesla ramps Semi, deepens $2.8bn AI chip pact

Tesla has begun volume production of its Semi electric truck, moving beyond pilot runs into commercial-scale output. The company placed the start of volume production in the third quarter of 2026 at its factory in Nevada. This marks a shift from earlier limited runs to a dedicated high-capacity program for the heavy truck platform. The Semi is part of Tesla’s broader push to pair vehicle manufacturing scale with software-driven systems.

Tesla indicated that the Nevada Semi facility has an installed annual capacity of 25,000 trucks. This installed capacity differs from a previously stated design capacity of up to 50,000 trucks per year, which described the maximum envisioned for the factory’s design. Neither figure represents actual quarterly production or delivery numbers, and Tesla did not disclose Semi production or delivery totals for the third quarter of 2026. The capacity data instead outline the potential scale the plant is built to handle as the program ramps.

In parallel with the Semi ramp, Tesla and Nvidia (NVDA) are jointly investing a combined KRW3.7 trillion, or about US$2.8 billion, to expand Samsung Electro-Mechanics’ AI chip substrate manufacturing capacity. The funding targets advanced substrates used in Tesla’s autonomous driving systems and Optimus humanoid robot hardware. The investment includes expansion of Ajinomoto Build-up Film substrate production facilities in South Korea, highlighting the strategic importance of substrate technology in the AI hardware supply chain.

A key focus of the partnership is Flip Chip-Ball Grid Array packaging, which connects chips to substrates through solder bumps and is viewed as a next bottleneck for AI components. Tesla and Nvidia have collectively paid Samsung US$2.8 billion to build capacity for this technology. Within this total, Nvidia is paying about US$2 billion, while Tesla’s share is approximately US$830 million. The investment helps Samsung develop its Sejong plant, supporting advanced packaging for AI-related chips.

Samsung’s Sejong facility, backed in part by Tesla’s contribution, is slated to complete construction and enter production in the second half of 2028. The plant is expected to begin supplying Tesla with Flip Chip-Ball Grid Array substrates in 2029. Samsung’s total investment in the facility is 4.27 trillion won, with Tesla covering 1.1 trillion won, or about US$830 million, of that amount. The project is designed to expand capacity for substrates critical to Tesla’s in-house chips and real-world AI systems.

Under the arrangement, Samsung has a five-year contract to supply Tesla with substrates for autonomous vehicles and humanoid robots. This priority supply is intended to support Tesla’s autonomous driving systems, its planned Cybercab service, and its Optimus humanoid robot platform. By securing long-term access to advanced substrates, Tesla aims to reduce dependence on broader commodity supply chains used by other automakers and AI hardware buyers. The agreement aligns component availability with Tesla’s roadmap for AI-intensive products.

The Semi ramp and AI substrate investments come as Tesla manages higher capital spending and margin pressures. For the third quarter of 2026, Tesla reported net income of US$787 million, down from US$1.37 billion a year earlier. Analysts have raised concerns about profitability as the company continues to fund autonomy, robotics, custom chips and energy projects. The new commitments to Samsung Electro-Mechanics add to this capital expenditure load, even as they support Tesla’s strategy of building reusable AI and power infrastructure.

Together, volume production of the Semi and the AI component investments deepen Tesla’s integration between manufacturing and advanced computing. The heavy truck platform extends Tesla’s vehicle portfolio into freight applications at scale, while the Samsung Electro-Mechanics partnership secures key hardware for software-centric offerings. These developments reinforce a business model that seeks to combine physical products such as trucks and robots with AI capabilities embedded in custom chips and supporting substrates.