Let us examine the data dispassionately. Tesla’s Optimus humanoid robot program has been positioned as a cornerstone of the company’s Master Plan Part IV—a vision to reimagine labor, mobility, and energy through AI-driven robotics 15. Yet the transition from concept to production flow reveals systematic bottlenecks that the current evidence suggests will constrain throughput for a prolonged period.
Timeline Slippage and Actual Output
The original goal of a production-ready design by 2023 was replaced by aspirations of “thousands in factories,” but the V3 (Gen 3) unveil has been pushed to 2026 13,17. Elon Musk has since cautioned that output “will be extremely slow at first” because “this is not like making a car” 13, and that the production rate is “literally impossible to predict” 13. This contrasts with an earlier target of 50,000–100,000 units in 2026 13, and a recent claim by VP Lars Moravy that production was “ahead of schedule” 13. However, the measured output data tell a different story: by early 2026, only hundreds of units had been built 2,3, and in January Musk acknowledged that no Optimus robots were performing useful work in factories 13. The removal of “volume production” language from Tesla’s first-quarter letter 6 signals a retreat from near-term scale.
Manufacturing Complexity and Infrastructure
Optimus comprises approximately 10,000 unique parts 13. Musk has described it as “the hardest product to scale manufacturing” Tesla has attempted because every component is new and no established supply chain exists 4,6,24. This necessitates building a supply chain from first principles, which compounds the learning curve. Concrete physical steps are underway: Tesla decommissioned the Model S and Model X lines at its Fremont factory and is installing first-generation Optimus production lines 10,13,21,22,23, with construction well advanced 5,12. Production is targeted to begin around late July or August 2026 13, and recent earnings calls confirmed that lines are being installed for a start “soon” 5,18,19,20,27. However, initial output will be directed toward training data collection and functionality development, not customer shipment 4,5.
Competitive Throughput Benchmarks
While Tesla establishes its production processes, competitors have already achieved meaningful scale. Unitree Robotics shipped more than 5,500 humanoid units in 2025 1,2,3, with humanoid revenue exceeding half of its nine-month total 17, and it received IPO approval in July 2026 2,3. AgiBot reached a production milestone of 10,000 general-purpose robots by March 2025, having ramped from 5,000 units in only three months 3,17. Agility Robotics is opening a 60,000-square-foot training facility, its Digit robot is generating revenue 9, and a reverse merger is expected to make it the first pure-play publicly traded humanoid company 9. These data points indicate that the competitive landscape is moving from prototyping to volume shipment.
Implications for Execution and Capital Allocation
For Tesla, Optimus represents a potentially transformative product—Musk has called it “the biggest product of all time” with the capacity to “end poverty” 16. Yet the chronic delays and candid assessments of manufacturing difficulty have contributed to investor disappointment 26 and elevated execution risk 26,29. Repurposing the Model S/X lines demonstrates top-down prioritization but sacrifices near-term automotive revenue, a trade-off whose return horizon remains unclear. Analyst consensus suggests that humanoid robotics products typically require a decade to mature 11. Even with Tesla’s resources, the absence of an existing supply chain 4 and the need for breakthroughs in dexterous manipulation—cited by some as the “last mile” of the humanoid story 11—make a rapid throughput ramp improbable.
Broader ecosystem developments show Tesla expanding battery cell production in Grünheide 7,8 and ramping the Semi and Megapack 5,14,20, which underscores its manufacturing breadth. The integration of Grok as a potential digital management layer for Optimus 4,25 hints at a vertically integrated AI-robotics stack. Nevertheless, the immediate binding constraint is physical: building a new robot without any inherited component infrastructure. The walk-back of firm timelines 3,17 and the deletion of volume production language 6 imply a protracted, halting ramp. Near-term sentiment will likely be determined by concrete evidence of progress at the Fremont line and further specificity from management 28.
