Tesla targets 10M Optimus robots yearly

Tesla doubles down on humanoids, aiming for 10 million Optimus robots a year.

Production data shows the company is already reorganizing its factory footprint to turn a robotics bet into a high-volume manufacturing operation. Tesla plans to begin Optimus production at Fremont in Q2 2026, with a first-generation line designed for about one million robots per year. Simultaneously, the company has begun site preparation for a second-generation, Texas-based facility intended to scale up to 10 million units annually over the long term. The shift comes as Tesla pivots from EV assembly to a robotics-first manufacturing ecosystem, backed by a push into vertical integration that includes an AI-focused hardware layer and a dedicated inference processor.

The fiscal frame matters. On an earnings call early this year, Tesla highlighted a cash-flow and margin backdrop that makes the robotics pivot less abrupt and more deliberate: $3.9 billion in operating cash flow and a 21% GAAP gross margin. The company framed Optimus as part of a broader automation strategy that could reshape how it approaches scale, supply chains, and feature development, all while attempting to maintain the velocity that has defined its EV business. Fremont’s first-gen line is not a throwaway demo; it’s engineered to prove the concept at a meaningful scale before the Texas facility stretches toward a 10x increase in annual output.

AI and software are not afterthoughts here. During the Robotics Summit & Expo discussions surrounding AI robotics, executives and engineers argued that AI is transforming how robots are programmed, learned, and deployed in production settings. The Shared theme: AI-driven systems can reduce initial programming intensity, letting teams teach tasks more fluidly and re-train as lines evolve. But the real-world deployment questions linger: how quickly can a robot learn new tasks, how much infrastructure—data networks, edge compute, and model management—will be required, and what does “re-learning” cost when process changes happen mid-shift? The dialogue underscored a practical truth Tesla will face: the software and AI stack must scale in lockstep with hardware to avoid bottlenecks that erase any cycle-time gains on the line.

From a practitioner’s perspective, the path to mass Optimus production is as much about execution as it is about concept. Integration teams report that the second-gen Texas plant will demand substantial floor space, power provisioning, and training hours for operators who will need to work with robots that can reconfigure themselves for different tasks. The questions aren’t just about fitment; they’re about sustainment—how long the lines will run without devoting constant engineering support, and how quickly maintenance cycles can be planned around unexpected downtime. Hidden costs tend to lurk in tool changes, security hardening, and data-management requirements that accompany AI-enabled automation.

Industry observers warn that the optimistic view—ten million bots per year—will require more than a grand plan. It will demand reliable onboarding of frontline teams, rigorous change management, and robust supply-chain discipline to deliver components at scale without crippling the line with stop-and-start outages. Yet the cadence of Tesla’s announcements signals a serious commitment to making robotics a core part of its manufacturing DNA, rather than a showpiece.

If the company can translate Fremont’s initial one-million-per-year milestone into Texas’ long-run ambition, the industry could see a rare blend of scale economics and practical, AI-assisted deployment proliferate on the factory floor. The next 18 months will tell whether the dream scales from a bold headline into an operational reality that materially shifts cycle times and throughput.

Sources

From EVs to robotics: Tesla targets 10M Optimus units with new Texas plant

AI robotics: Moving from the lab to the real-world factory floor