Tesla Plans Massive “Terafab” for AI Chips, Considers Intel Partnership

Elon Musk has made it clear: Tesla isn’t just in the car business anymore. It’s in the intelligence business.

At Tesla’s annual shareholder meeting, Musk revealed plans for what he called a “gigantic chip fab”, a facility that could become one of the largest semiconductor manufacturing sites on the planet. The goal is simple: meet the skyrocketing demand for Tesla’s custom-designed AI chips, which power everything from Full Self-Driving (FSD) to the company’s upcoming autonomous “Cybercab” fleet.

“You know, maybe we’ll do something with Intel,” Musk told shareholders, sparking cheers and immediate speculation about a new partnership. “We haven’t signed any deal, but it’s probably worth having discussions with Intel.”

That casual remark was enough to move markets. Intel’s stock jumped nearly 4% in after-hours trading, a rare win for the chipmaker that’s been struggling to keep pace with Nvidia’s dominance in AI hardware. With the U.S. government recently taking a 10% stake in Intel to bolster domestic chip manufacturing, a Tesla collaboration could be the opportunity it’s been waiting for.

Meanwhile, Tesla’s chip roadmap continues to accelerate. Musk confirmed that the company’s AI5 chip, its fifth-generation self-driving processor will enter limited production in 2026, ramping to full-scale output in 2027. A follow-up AI6 chip is already planned, using the same fabs but offering roughly double the performance and efficiency by mid-2028.

Still, Musk said that even top-tier partners like TSMC and Samsung won’t be able to meet Tesla’s future demand for AI silicon. That’s where the “Terafab” idea comes in, a facility capable of producing over 100,000 wafer starts per month, roughly on par with some of the world’s largest semiconductor foundries.

“I can’t see any other way to get to the volume of chips we’re looking for,” Musk said. “So I think we’re probably going to have to build a gigantic chip fab. It’s got to be done.”

The proposed fab would focus on Tesla’s in-house AI chips, which Musk claims will be significantly cheaper and more energy-efficient than Nvidia’s flagship Blackwell GPU. According to Musk, Tesla’s design could consume just a third of the power while costing roughly one-tenth to produce. A bold claim, but one that fits Tesla’s history of rewriting industry economics through vertical integration.

Building a semiconductor plant, however, isn’t as simple as stamping out Model Ys. Chip fabs are among the most complex and capital-intensive facilities in the world, often requiring years of development and billions in investment. Musk didn’t share details about the location or construction timeline, but given Tesla’s recent success with Giga Texas, a U.S.-based Terafab wouldn’t be a surprise.

Musk also suggested the new chips would be deeply optimized for Tesla’s own software stack, reducing reliance on third-party hardware and improving long-term cost control. By owning the full pipeline from software to silicon, Tesla could fine-tune every element of its AI systems, pushing faster FSD updates and improving vehicle response times on the road.

The broader implication? Tesla could soon compete not just with automakers, but with the likes of Nvidia, AMD, and Intel themselves. Musk has repeatedly said Tesla’s future lies in AI, robotics, and energy, and this move cements that evolution.

Investors clearly approve. Shareholders just voted to reinstate Musk’s record-setting $1 trillion compensation plan, signaling strong faith in his push to turn Tesla into an AI and robotics leader.

For Tesla owners and enthusiasts, this means more than faster chips, it means a smarter ecosystem. Every vehicle could eventually become a rolling supercomputer, trained on billions of miles of real-world data and powered by hardware Tesla builds itself.

Source: Reuters