Elon Musk Announces Ambitious Plan to Outproduce All AI Chip Makers Combined with Tesla's Custom Silicon

Elon Musk claims he will 'build chips at higher volumes ultimately than all other AI chips combined' -- Tesla AI engineering team has AI5 chip ready to go and is setting its sights on AI6

Hopefully the goal isn't as far away as Tesla's Full Self Driving. Elon Musk isn't one to shy away from grand promises, and the latest one involves AI chips. The entrepreneur has reminded people (and investors) via an X post that Tesla has had an "advanced AI chip and board engineering team" for several years now. He goes on to state that the chips coming out of this project will make Tesla "the leader in real-world AI," as bold a claim as any, and that the firm plans to make more AI chips than all other chipmakers, combined. Musk claims this engineering team has already deployed millions of "AI chips" in cars and data centers. The keyword here is "cars," however. The actual count of deployed hardware would be hard to pin down, as the statement implies that 2016-vintage Nvidia Drive PX 2 SoCs inside early Teslas count as much as a latest-gen Blackwell data center accelerator. The technologist goes on to specify that Tesla currently uses its own AI4 chips in its cars and is close to taping out AI5. Going a step further, he states it expects Tesla to have a new iteration ready every 12 months, but more importantly, that it expects to "build chips at higher volumes ultimately than all other AI chips combined," a statement that merits some healthy skepticism, given how many chips Nvidia, AMD, Intel, Qualcomm, et al currently produce. Tesla is currently a customer of TSMC and Samsung, and has expressed interest in working with Intel's foundries, given the current situation with worldwide chip supply constraints. Musk is considering building his own TeraFab chip factory, too. The design schedule also seems quite aggressive, but not impossible, given how much money has been poured into neural network accelerator chip design lately. As befits a grand presentation, Musk goes on to connect the company's AI developments to the Optimus robotics project, claiming the new chips will "save millions of lives" thanks to advanced medical care via unspecified integration with Tesla's Optimus project. Optimus robots were demonstrated in 2022, but they were being teleoperated, as the company has yet to produce an actual autonomous bipedal humanoid. The predictions about Optimus' abilities can be arguably quite optimistic, as well (pun unintended). In 2024, Musk stated the Optimus robots would be working in Tesla factories this year, and as of this writing, no public demonstrations have been made about these capabilities. In fact, just last summer, it seemed that Optimus robot production came to a grinding halt. Having said that, it's worth keeping in mind that Elon Musk also owns xAI, a company with actual released and working products, namely the Grok chatbot, the Aurora image generator, and DeepSearch engines. It should be a safe enough assumption that Tesla and xAI are sharing technology, perhaps somewhat enabling Musk's ideals of assistant doctor robots. It's a safe bet that such a feat will take a while, as Tesla's Full Self Driving is now marking its ninth year since it was promised, and fifth since it entered the Beta stage.

Elon Musk Claims Tesla Must Outproduce NVIDIA, AMD, and All Other AI Chip Makers in Volume if It Wants to Win the Real-World AI Race

Well, Elon has apparently 'doubled down' on his ambitions, as he envisions producing chips at a pace that surpasses all other tech giants out there, to back Tesla's AI-focused plans. Tesla's CEO has made really interesting statements around chip buildouts in the past few weeks, and it appears that the automaker is eager to scale up production capacities to meet the estimated demand. At a post on X, Elon has revealed that Tesla has been involved in building AI chips for several years now, and the firm is now operating on an annual product cadence, releasing new platforms after "every 12 months". And, more interestingly, Tesla's CEO aims to produce AI chips at a volume higher than that of every other manufacturer, indicating that the scale of output would be unprecedented, to say the least. Well, this isn't the first time Elon has expressed his desire to manufacture chips at a gigantic scale, since, based on what he had said previously, Musk wants to produce up to a whopping 200 billion AI chips per year, driven by the adoption of Tesla's FSD-focused fleet, but we still don't know how such a prominent figure would be integrated, unless Elon has other plans. We do know that Tesla is expanding into newer frontiers, with Optimus and Cybercab, but even after that, such a semiconductor demand not only seems far-fetched but also impossible to produce with existing resources. Speaking of production constraints, Elon has revealed 'optimistic plans' to address this bottleneck as well, expressing a desire for a 'TeraFab' buildout that would fulfill the semiconductor demand Tesla anticipates. Moreover, Tesla has diversified its semiconductor sourcing strategy, integrating TSMC and Samsung for AI5 and next-gen AI6 chips, and simultaneously plans to include Intel Foundry in the mix as well. Even then, Elon believes that chipmakers cannot fulfill his demand at all, which is why his plans around a chip supply buildout are sensible, according to him. We have dived into the complexities of building a chip supply chain earlier as well, but to sum it all up, it isn't an easy job, especially for an entity like Tesla, which has little experience in semiconductor manufacturing. Elon has been building up 'bits' of a chip supply chain for SpaceX, as we reported earlier. However, to meet the demand Tesla estimates, the only way forward for the firm is through collaboration.