AustinIO > Giga Austin Nexus > Optimus Humanoid Production

Optimus Production at Tesla Giga Texas

Optimus is the program at Giga Austin with the largest theoretical scale and the smallest current output.

Tesla's Q1 2026 Form 8-K confirmed both ends of that gap explicitly. The company is preparing the Texas site for a long-term annual production capacity of 10 million humanoid robots while simultaneously running the first-generation pilot line at Fremont, California, on the converted Model S and Model X assembly bays at a target of 1 million units per year.

The Giga Austin Optimus factory broke ground in the first quarter of 2026 on the North Campus expansion adjacent to the existing 10-million-square-foot vehicle factory. Geopier ground-reinforcement equipment was first observed by drone in early April 2026, with major structural work targeted to complete by end of 2026.

Permit documents reviewed by industry observers indicate Tesla is seeking to add over 5.2 million square feet of new building space to the Giga Texas North Campus by end of 2026 — a footprint that includes the Optimus factory alongside the Terafab pilot fab and other adjacent buildings — at an estimated construction investment of $5 billion to $10 billion.

Low-volume Optimus output from the Austin facility is projected to begin as early as summer 2026, with the full volume ramp targeting 2027. The Austin program is the second-generation Optimus production line; Fremont is generation one.

The Two-Site Production Architecture

Tesla's stated Optimus manufacturing strategy splits the program across two sites with distinct roles.

Site	Generation	Site Type	Annual Capacity Target	Strategic Role
Fremont, California	Generation 1	Retrofit (converted Model S/X assembly bays)	1 million units	Production-engineering proving ground
Giga Austin (North Campus)	Generation 2	Purpose-built (clean-sheet on undeveloped land)	10 million units (long-term design target)	Volume production at engineered scale

Fremont serves as the production-engineering proving ground. It is the site where Tesla learns to manufacture a humanoid at scale before committing to the larger Texas footprint. Production of Model S and Model X is ending in Q2 2026 specifically to free the assembly bay floor space for Optimus.

Giga Austin hosts the second-generation line, designed from the start for 10x the Fremont throughput. The Austin site does not retrofit existing space the way Fremont does; it is purpose-built on previously undeveloped land on the North Campus, allowing the production architecture to be optimized for the robot rather than constrained by an existing building.

The split mirrors the manufacturing-doctrine pattern Tesla has used before: prove a process at smaller scale on existing capacity, then build a clean-sheet large-scale facility once the learning curve has flattened.

The 2026-2027 transition window is when both lines run concurrently — Fremont ramping toward its 1M target while Austin moves from groundbreaking through structural completion to equipment installation and pre-production validation.

What Optimus Is, Mechanically

Optimus V3 — also referred to as Gen 3 and the production-intent version — is the first iteration designed from the start for high-volume manufacturing rather than for prototype demonstration. The body retains the Gen 2 form factor of approximately 1.73 meters tall and 57 kilograms. The transformative change is in the hands.

Specification	Optimus V3
Height	~1.73 meters
Mass	~57 kilograms
Total degrees of freedom	~72 (28 body + 22 per hand × 2)
Hand DoF	22 per hand
Actuators per forearm/hand assembly	25 (50 total across both arms)
Hand actuation method	Tendon-driven cables; actuators in forearm, not hand
Total unique components per robot	~10,000
On-board AI compute	AI4 → AI5 transition (~40x AI4 inference performance)
Battery	~2.3 kWh (8 hr light-duty / 4-6 hr intensive operation)
Natural language	xAI Grok LLM

Tesla filed four international patents on the forearm, wrist, joint, and hand architecture in April 2026 — the first detailed public mechanical blueprint of the production design.

Musk has stated the hand represents approximately 60 percent of the engineering difficulty of the entire Optimus program, harder than the Cybertruck, with the production-intent design only reaching maturity in early 2026.

Most components have no existing supply chain. Tesla is building the majority of the actuator stack, the harmonic and planetary-roller-screw drives, and the joint assemblies in-house because no commercial supply base exists at the cost and quality required.

The Semiconductor Density Inside Each Robot

Optimus is one of the most semiconductor-intensive embodied AI systems ever conceived. Each unit contains an estimated 1,100 to 2,200 individual semiconductor devices distributed across nine major categories.

Category	Function
Logic and compute	Main inference SoC, sub-system controllers
Memory	DRAM, flash, embedded memory
Power semiconductors (largest count)	SiC and GaN MOSFETs for actuator drive and DC-DC conversion (400-800 per robot)
Sensors	Tactile, force/torque, IMU, position, temperature, voltage, current
Analog and mixed-signal	Signal conditioning, ADC/DAC, amplifiers
Embedded microcontrollers	Joint-level motor control, peripheral subsystems
RF and networking	Wi-Fi, cellular, internal data fabric
Optoelectronics	Cameras, LiDAR (where present), display drivers
Security silicon	Secure boot, cryptographic acceleration, OTA integrity

Power semiconductors are the largest single category by device count at 400 to 800 devices per robot. Each of the 80-plus joints requires three-phase motor drive circuitry. Each motor requires voltage, current, speed, torque, position, and temperature sensing. Each finger and toe carries at minimum a tactile pressure sensor. Each battery cell carries voltage monitoring with multiple temperature sensors throughout the pack.

The compounding effect at scale is the structurally important number. At the 10-million-unit Austin annual capacity Tesla has stated, Optimus production alone would consume 4 to 8 billion power semiconductor devices per year — a demand signal that exceeds current global SiC and GaN production combined.

This is the central reason the Optimus factory's siting at Giga Austin matters strategically. The Terafab pilot fab on the same North Campus, designed to produce Tesla's AI5 logic and adjacent silicon for vehicle and humanoid programs, is the company's hedge against being supply-constrained on its own demand.

Detailed semiconductor stack coverage is available at SemiconductorX: Spotlight on Humanoids & Robotics ↗.

The Fremont Pilot Line as Production Doctrine

The Fremont line is operationally significant beyond its 1-million-unit capacity because it is where the Optimus manufacturing doctrine is being established.

Tesla's Q4 2025 earnings disclosure and subsequent Q1 2026 reporting indicate the company is converting the Model S and Model X assembly bays — among the most experienced production teams in the company — to humanoid manufacturing.

The explicit logic is that Optimus production is closer to vehicle assembly than to any existing precedent, and that the institutional knowledge from twenty-plus years of Tesla automotive production is the relevant starting point.

Musk has stated that one Optimus, at full operational capability, equals approximately 5x human productivity at lower cost, running 24/7 without breaks if tethered to a power source. The recursive implication — Optimus robots manufacturing additional Optimus robots — is the long-term efficiency claim Tesla has made for the program.

That said, Musk acknowledged on the Q4 2025 earnings call that current Optimus units are "still very much in the R&D phase" and not yet performing useful work at scale.

The Fremont line's primary near-term output is intended to be internal Tesla deployment. Thousands of units will be placed on Tesla's own factory floors through 2026 as a real-world testing environment that surfaces failure modes, actuator wear patterns, software edge cases, and integration challenges before external sale.

The internal-deployment-first strategy is also the cost-reduction strategy. Tesla's stated long-term Optimus selling price target is $20,000 to $30,000.

Cost Category	Estimated Share of Current BOM
Actuators (50 per robot)	~56%
All other components	~44%

Driving total cost down to the $20-30K selling price target requires the same vertical integration and production-scale playbook that made Tesla vehicles progressively more affordable, executed against a robot whose component complexity exceeds the Model Y by an order of magnitude.

The Austin Buildout and What 5.2 Million Square Feet Implies

The North Campus expansion at Giga Austin is the largest single industrial buildout in Tesla's history when measured by floor area.

Permit documents indicate over 5.2 million square feet of new building space targeting end of 2026 — the equivalent of more than 90 football fields of additional roofed manufacturing capacity adjacent to the existing 10-million-square-foot main factory.

The Optimus factory is one of multiple buildings on the expansion footprint, alongside the Terafab semiconductor pilot fab (which broke ground April 22, 2026), a Megapack production facility, additional Cortex AI compute halls, and supporting infrastructure.

The aggressiveness of the build timeline is the analytically significant feature. 5.2 million square feet of new construction in approximately one calendar year is a build cadence that compresses what would normally be a multi-year industrial buildout into a single annual cycle. The scale of investment is estimated at $5 billion to $10 billion across the North Campus expansion.

The 10-million-unit annual Optimus production target is unprecedented in humanoid robotics.

Operator	Stated Production Target	Component Sourcing
Tesla (Optimus)	10 million units/year (Giga Austin design capacity)	Custom precision components, mostly in-house
XPeng (Iron)	1 million units/year by 2030	Car-grade rather than custom precision

Whether Tesla can compress humanoid production scale-up into the 2026-2027 window the company has projected, against a supply chain that does not yet exist for the precision actuators and sensors the robot requires, is the central operational question for the Austin program.

The bet Tesla is making with the North Campus footprint and the in-progress Terafab is that vertical integration, achieved fast enough, becomes a competitive moat that no other operator can replicate — and that the same recursive feedback that defines the EV production stack at Giga Austin will compound at the humanoid level as well.

Production Capacity Snapshot (Q1 2026)

Site	Generation / Status	Annual Capacity Target
Fremont, California	First-generation line; converting from Model S/X assembly bays; under construction; ramp targeting late 2026	1 million units
Giga Austin (North Campus)	Second-generation line; foundation work underway Q2 2026; structural completion targeting late 2026; volume ramp targeting 2027	10 million units (long-term design capacity)

Capacity figures are from Tesla's Q1 2026 Form 8-K filing. The Austin 10-million-unit figure is the long-term design target rather than installed capacity at any specific year — it represents the production scale the building is being engineered to support, not the rate at which units will leave the line in 2027.

Outlook

The Optimus program is the bet at Giga Austin with the longest payoff timeline and the largest potential revenue redirection if it succeeds.

Musk's compensation package approved by Tesla shareholders in November 2025 ties a substantial portion of his award to Optimus delivery milestones, with one million humanoid robots delivered by 2035 as an explicit target.

The 2026-2027 questions the program will resolve are operational rather than visionary:

Whether the V3 hand reaches manufacturing maturity at the cost the financial model requires. Whether internal Tesla factory deployment validates the productivity claims at scale. Whether the actuator and precision-component supply chain can be built fast enough to support the Austin ramp. Whether the recursive feedback between Optimus production and Optimus deployment compounds the way Tesla's manufacturing doctrine projects.

The semiconductor density question is the substrate-level question. At 10 million Optimus units per year requiring 400-800 power semiconductor devices each, the program's success is materially dependent on the Terafab pilot fab and broader semiconductor capacity buildout that the rest of the Giga Austin North Campus is being constructed to support.

The Optimus factory is not analyzable as a standalone industrial site. It is analyzable as one program in the four-program campus concentration, drawing substrate from and feeding substrate to the others.

The 5.2 million square feet of new construction underway is the visible manifestation of Tesla's bet that the four-program integration produces compounding advantages that justify the capital concentration.

Related Coverage