Tesla Is Building a Massive Optimus Factory at Giga Texas, Targeting 10 Million Robots Per Year
6 min read read
Tesla's ambition for its Optimus humanoid robot has always been measured in numbers that strain credulity. During the Q1 2026 earnings call, the company added specificity: it has begun site preparation for a dedicated Optimus manufacturing facility at Giga Texas, is seeking permits for 5.2 million square feet of new construction on the campus's North expansion, and is targeting a long-term production capacity of 10 million robots per year at that single location.
For context, Toyota — the world's largest automaker by volume — produces roughly 10.5 million vehicles per year across its entire global manufacturing network. Tesla is planning to match that output with one product, at one location.
Two Factories, Two Timelines
Tesla's Optimus manufacturing strategy splits across two facilities with different timelines and different generation targets.
The nearer-term operation is at Fremont, California: first-generation Optimus production lines are underway, with a pilot target of 1 million units per year. Tesla confirmed during Q1 earnings that these lines are "well underway" with production expected to begin in Q2 2026. The Fremont facility serves as Tesla's proving ground for manufacturing processes before scaling to Texas.
The larger-scale facility is the Giga Texas North Campus expansion. Permit documents filed with Texas authorities show Tesla seeking to add 5.2 million square feet of new building space to its existing 10-million-square-foot, 2,500-acre campus in Austin. The estimated construction investment is between $5 billion and $10 billion. This facility is designed around Optimus V4 and carries the 10 million units per year long-term production target.
| Facility | Location | Target Capacity | Timeline | Generation |
|---|---|---|---|---|
| Fremont Factory | Fremont, CA | 1M units/year | Q2 2026 start | Gen 1 (current) |
| Giga Texas North | Austin, TX | 10M units/year | 2027+ ramp | V4 (future) |
Why Texas Makes Strategic Sense
The decision to build the high-volume Optimus factory in Texas rather than California reflects a convergence of infrastructure, cost, and AI architecture considerations.
Giga Texas already hosts two of Tesla's largest AI training clusters: Cortex 1 and Cortex 2, collectively totaling more than 230,000 H100-equivalent GPUs. Training the software that will run inside Optimus units — in the field, in real time — and the factory that will produce those units will share the same campus. The feedback loop between production hardware and training infrastructure is shorter when they're co-located.
"The proximity to Tesla's AI training cluster Cortex 1 and 2, both located at Giga Texas and now totaling over 230,000 H100 equivalent GPUs, means the Optimus software stack and the factory producing the hardware will share the same campus." -- The Robot Report, April 2026
Additionally, in April 2026, Tesla completed the tape-out of its AI5 chip — an inference processor designed specifically to power Optimus units in the field. Unlike the training chips housed in Cortex, AI5 is meant to run inside each robot's body, performing real-time perception, decision-making, and motor control. Completing tape-out is a critical milestone: it means the chip design is frozen and silicon production can begin.
The Scale Challenge
Building 10 million robots annually requires solving manufacturing problems that don't yet have established solutions. A humanoid robot contains 28 degrees of freedom in its hands alone, precision actuators throughout its joints, and a sensor suite that includes cameras, microphones, and force sensors in its fingertips. The bill of materials per unit is estimated at $20,000 to $30,000 at consumer price points — which implies a manufacturing cost that must come significantly lower for healthy margins.
Tesla's advantage, to the extent it exists, is that it has already solved analogous scaling problems. Building 1.8 million electric vehicles per year required mastering precision battery cell production, high-volume die casting, and automated assembly at a scale that defeated other automakers' initial attempts to compete. The 4680 battery cell program — plagued by early yield problems — eventually hit cost targets through relentless process iteration.
Revenue Potential at Scale
Musk has publicly targeted a retail price of $20,000 to $30,000 per unit for consumer Optimus, with public sales expected by end of 2027. At 10 million units annually and a $25,000 average selling price, the theoretical revenue figure approaches $250 billion per year — roughly ten times Tesla's current automotive revenue.
| Scenario | Annual Units | ASP | Revenue Potential |
|---|---|---|---|
| Fremont pilot (near-term) | 1,000,000 | $25,000 | $25B |
| Texas full ramp (long-term) | 10,000,000 | $25,000 | $250B |
| Tesla automotive (2025 actual) | 1,789,226 | ~$43,000 | ~$77B |
Those figures remain hypothetical. Production hasn't started at Fremont yet, the Texas factory is still in site preparation, and the AI5 chip only completed tape-out in April. The gap between a tape-out and a chip inside a mass-produced robot is measured in years of manufacturing ramp.
The Bottom Line for Tesla Watchers
Tesla's Optimus ambition has shifted from roadmap slide to active construction permit. The Fremont pilot line — modest at 1 million units per year — gives Tesla a proving ground for the processes and supply chains that will eventually need to scale tenfold at Giga Texas. Whether a company that makes electric cars can also become the world's largest robot manufacturer remains an open question, but the construction permits, the AI5 tape-out, and 230,000 GPUs worth of training infrastructure suggest Tesla is spending real money to find out.
Photo: Industrial manufacturing facility / Pexels