Elon Musk is plotting Tesla’s next major leap beyond the automotive sector: building its own “Terrafab” semiconductor fabrication facility to fuel the company’s expanding AI ambitions.
During Tesla’s 2025 Annual Shareholder Meeting, Musk stated that global supply from partners like TSMC and Samsung may fall short of meeting Tesla’s growing demand for AI chips. He proposed constructing a “massive chip fabrication plant” to secure Tesla’s AI chip supply, identifying semiconductors and power as the two primary constraints on the company’s future growth.
Bringing Chip Manufacturing In-House
Musk emphasized that Tesla’s long-term strategy involves controlling every aspect of its AI hardware—from design to production. He warned shareholders that reliance on external foundries could hinder the company’s expansion into robotics and automated systems.
While Tesla currently partners with TSMC and Samsung to manufacture its custom-designed chips, it may pursue deeper collaborations or entirely new pathways. Musk confirmed discussions with Intel about a potential partnership but clarified that no agreements have been finalized.
“Engaging in talks with Intel could be worthwhile,” he noted, adding that Tesla’s top priority is ensuring sufficient chip supply to meet future demand. Musk revealed that Tesla has already secured the full output from one partner’s fab—but even that might not be enough to satisfy projected needs.
Chips Promising Major Gains in Power Efficiency and Performance
Musk announced that Tesla’s in-house AI5 processor is nearing production, with mass manufacturing set to follow initial runs. Designed specifically for the company’s automation and robotics initiatives, the chip is optimized to run Tesla’s neural networks more efficiently than current hardware.
According to Musk, the AI5 consumes roughly one-third the power of NVIDIA’s Blackwell architecture and costs less than one-tenth as much. Its integer-focused design aligns closely with Tesla’s AI software stack, delivering high performance while drastically cutting energy use.
Within a year of AI5 entering production, the next-generation AI6 chip is expected to launch, offering approximately double the performance.
Giga… But Much Bigger
Musk outlined plans for the proposed Terrafab to start with a target capacity of 100,000 wafers per month, with the potential to scale up tenfold as Tesla’s AI footprint grows.
He explained to shareholders that this output level is essential to support Tesla’s expanding ecosystem of autonomous vehicles and humanoid robots. Without such capacity, he warned, the company risks hitting a chip supply bottleneck just as demand for its AI systems accelerates. If external suppliers can’t keep pace, Musk added, building an exceptionally large in-house fab becomes the only viable option.
Dubbing the initiative “Giga, but bigger,” Musk positioned Terrafab as the industrial-scale platform needed to realize Tesla’s AI vision.
Tesla’s Future Is Built on Chips
Musk told shareholders he has become “super hardcore” about semiconductors, underscoring their central role in shaping Tesla’s trajectory. He reiterated that chips and electricity are the two ultimate factors determining how fast the company can scale.
He stressed that this technology extends far beyond powering vehicles. Tesla’s AI chips will also drive its Optimus humanoid robots, autonomous vehicle fleets, and large-scale inference infrastructure—forming the backbone of its broader AI strategy.
By deepening its investment in silicon, Musk said, Tesla aims to secure the computational foundation necessary to sustain its next wave of AI and robotics growth.
