Tesla may build giant chip factory for AI cars and robots

Close-up portrait of Elon Musk in a bright cleanroom with towering semiconductor fabrication equipment behind him, a giant iridescent silicon wafer glowing like a halo, sleek outlines of an autonomous Tesla car and a humanoid robot emerging in the background, a glossy red Tesla logo emblem floating center-right, high contrast warm reds against cool cyans, crisp editorial style, shallow depth of field, dramatic yet neutral tone

Tesla may build its own large semiconductor factory to meet growing chip demands as it expands in artificial intelligence and robotics. Elon Musk told shareholders the company would likely need a massive facility capable of one million wafer starts per month over time, beginning at around 100,000. He made the comments after investors approved his $1 trillion compensation plan at the annual meeting in Texas.

Scaling Production and Custom Chips

Musk said electricity and semiconductor supply are the key constraints to Tesla’s ambitions. Those ambitions include scaling vehicle output and accelerating production of its Optimus humanoid robot. According to Just Auto, Musk described the shift as opening a new book for Tesla, not just a new chapter.

The newly approved compensation package will vest in 12 parts tied to operational and financial targets. Those targets include delivering 20 million vehicles, reaching 10 million active Full Self-Driving subscriptions, putting one million Optimus units into customers‘ hands, and deploying one million robotaxis.

AI5 Processor Performance

Musk said Tesla’s custom AI5 processor uses a third of the power of Nvidia’s Blackwell chip while costing less than 10 percent as much. The chip is optimized for Tesla’s AI software stack and is not meant to be a general-purpose processor. It is designed to run the company’s vehicles and robots.

Supplier Talks and China Approval

Tesla currently relies on Taiwan Semiconductor Manufacturing Company and Samsung Electronics to make its AI chips at facilities in Asia and the US. Musk added the company is assessing a possible arrangement with Intel’s contract manufacturing arm. He said Tesla might do something with Intel but has not signed any deal.

Even with multiple suppliers, Musk cautioned output could fall short of Tesla’s needs. He said the company will take as many chips as suppliers can provide, but beyond that, Tesla may have to make them itself or get stuck. TSMC’s annual capacity in 2024 reached about 17 million 12-inch equivalent wafers, underscoring the scale of Musk’s proposed target.

Progress also depends on autonomous driving approval in China. Tesla delivered 1.79 million vehicles in 2024, with more than a third in China. The company has yet to secure approval to launch Full Self-Driving there. Musk said he hopes for full approval around February or March, based on what regulators have told Tesla.

Total
0
Shares
Previous Post
Editorial photo montage with medium-close portraits of Sam Altman and Sarah Friar facing inward, a clean OpenAI logo centered between them, subtle US Capitol dome silhouette and stylized chip patterns in the background, bright teal-to-amber gradient lighting with high contrast and glossy highlights, neutral expressions, crisp professional style, no text

OpenAI executives deny wanting taxpayer money for AI chips

Next Post
Neutral close-up portrait of Sam Altman centered with a softly blurred backdrop of towering futuristic data center racks, a subtle OpenAI logo halo behind him, hints of gold coin stacks in the foreground for scale contrast, bright and clean editorial look with vivid teal and electric blue against warm amber highlights, tight framing, no text

OpenAI needs to match Google’s sales in 4 years to pay bills

Related Posts