Tesla is accelerating its artificial intelligence strategy with the upcoming AI5 chip, which will be produced by two major semiconductor manufacturers—Samsung in Texas and TSMC in Arizona—according to CEO Elon Musk during the company’s third-quarter earnings call. This dual-foundry approach reflects Tesla’s growing emphasis on securing sufficient supply to fuel both its electric vehicles and emerging AI-driven projects.

Musk emphasized that Tesla’s objective is not merely to meet current demand, but to intentionally create a surplus. “Our explicit goal is to have an oversupply of AI5 chips,” he said, noting that any chips not immediately required for vehicles or robotics applications will be deployed in Tesla’s expanding data center infrastructure.

Since moving away from Nvidia’s Drive platform in 2019, Tesla has focused heavily on building proprietary hardware to power its autonomous driving and robotic projects. However, Musk confirmed that Nvidia’s GPUs remain essential for training Tesla’s AI models, especially given Nvidia’s dominant role in the high-performance AI computing market. “We’re not about to replace Nvidia, to be clear, but we do use both in combination,” he noted. Tesla also revealed it now has computational resources equivalent to approximately 81,000 Nvidia H100 chips, underscoring the scale of its AI operations.

The development of the AI5 chip provides insight into Tesla’s evolving semiconductor roadmap after the departure of former Apple engineer Peter Bannon earlier in the year. Bannon had been instrumental in leading Tesla’s in-house chip initiatives and spearheaded the Dojo supercomputer project, which is intended to enhance full self-driving capabilities by improving neural network training efficiency.

Initially announced at Tesla’s 2024 shareholder meeting, the AI5 chip represents the next-generation Autopilot hardware that processes real-time signals required for autonomous navigation. Its production became more visible after Samsung disclosed a $16.5 billion contract with an undisclosed buyer—later confirmed by Musk to be Tesla. While Musk previously said that TSMC would manufacture AI5 and Samsung would handle its successor, the AI6, he clarified during the recent call that both foundries will now produce the AI5 chip within their U.S. facilities.

Musk described the AI5 as being built on a “half reticle” design, meaning it is roughly half the size of full reticle chips commonly used in high-end AI hardware by companies like Nvidia and AMD. This smaller form factor suggests Tesla is prioritizing cost-efficiency, power optimization, and more specialized use cases tailored to its vehicles and robotics platforms, rather than competing directly with broader general-purpose AI chips.

The race to develop custom AI chips is becoming increasingly competitive across the tech landscape. Hyperscale cloud providers such as Google, Amazon, and Microsoft have accelerated investments in proprietary solutions, seeking not only to diversify away from Nvidia’s architectures but also to optimize performance for their specific workloads. Analysts widely believe that custom-designed chips can potentially outperform generalized processors in focused applications, particularly when integrated deeply with a company’s existing ecosystem.

Among consumer-focused technology firms, Apple remains the only major company that fully designs and deploys its own silicon in both hardware products and data centers. Tesla is now positioning itself alongside Apple by engineering AI chips designed exclusively for its ecosystem. Musk noted that Tesla’s singular-customer focus significantly streamlines the design process. “Tesla only has to satisfy requirements from one customer,” he said. “That makes the design job radically easier and means we can delete a lot of complexity from the chip.”

According to Musk, Tesla’s engineering team has removed legacy elements such as GPUs and traditional signal processors from the AI5, replacing them with highly specialized components tailored for autonomous decision-making and edge AI processing. He predicted that the AI5 may deliver superior performance-to-cost metrics compared to industry-standard AI chips. “The chip could have the best performance per dollar for AI—maybe by a factor of 10,” he stated.

While praising Nvidia’s ability to meet a wide spectrum of complex industrial demands, Musk emphasized Tesla’s contrasting pursuit of focused efficiency. “Nvidia has done an amazing job of dealing with almost an impossibly difficult set of requirements,” he acknowledged. “But in our case, we are waiting for radical simplicity.”

Musk’s AI ambitions extend beyond Tesla’s automotive division. His AI startup, xAI, which collaborates closely with Tesla, has emerged as a major purchaser of Nvidia hardware. xAI is currently constructing a massive supercomputing campus in Memphis, Tennessee, which represents its second large-scale facility in the area. The infrastructure will be built around Nvidia’s Grace Blackwell chips and is expected to support both Tesla’s autonomous driving research and xAI’s broader generative AI development.

Tesla’s decision to secure dual fabrication partnerships with Samsung and TSMC highlights a broader strategy to safeguard supply and mitigate risk amid growing competition for advanced AI chip production capacity. With increasing global demand for high-efficiency AI silicon, Tesla is aiming to ensure consistent output as it scales self-driving initiatives, humanoid robotics, and AI-driven optimization for its energy storage systems.

As Tesla moves toward mass deployment of its Full Self-Driving capabilities and further integration of AI across its vehicle lineup, the AI5 chip could play a key role in differentiating the company’s products and improving cost structures. Meanwhile, Tesla’s presence in the AI infrastructure ecosystem signals a future in which the company becomes not only a vehicle manufacturer but also a central player in specialized AI computing.

With the AI5 marking the next chapter in Tesla’s silicon journey, industry observers are closely watching how the company balances its reliance on Nvidia while scaling its proprietary computing platforms. The success of the AI5 chip may determine how aggressively Tesla will push forward with AI6 and future iterations to redefine its role in both the automotive and AI technology sectors.