Tesla Optimus Claim Verification Tracker
In our humanoid robotics market deck, you will find everything you need to understand the market
Summary
Out of the 11 Optimus promises we tracked from Elon Musk between 2021 and 2025, only one has actually been fulfilled: the August 2021 promise to show a working prototype the following year. The other ten are either still unproven, only partially shown, or outright missed.
This tracker covers everything Musk has said publicly about Optimus through April 18, 2026. His most recent real claim goes back to March 2025, so the gap between his last specific promise and today is already more than a year long.
The clearest broken promise is the April 2024 commitment that Optimus would handle factory tasks by the end of 2024. More than 15 months later, there is still no public proof this ever happened.
Musk has also changed his 2025 Optimus production target at least two or three times in quick succession. The number moved from "several thousand" in July 2024, to "roughly 10,000" in January 2025, down to about 5,000 baseline in March 2025. Public estimates suggest Tesla actually built fewer than 500 units in 2025.
None of the favorable Optimus demos we found happened in a real customer or live production environment. Everything credible sits on a stage, inside a lab, inside a teleoperated setup, or inside a controlled Tesla facility. Several of the most impressive 2024 hand demos were later confirmed by Bloomberg to be remotely operated by humans.
There is also zero publicly available productivity data for Optimus. No uptime, no cycle time, no throughput, no ROI. Tesla's own January 2026 deck still describes Gen 3 as the design that will eventually go into mass production, which is not the language of a program already working at scale.
Our takeaway is simple. Musk proved the prototype exists. He has not yet proven that the robot is useful, affordable, or produced at scale.
Tesla Optimus Claim Verification Tracker
This table brings together every first-hand public statement Elon Musk has made about Tesla Optimus since 2021, covering capabilities, pricing, timelines, and production targets. We grouped repeated statements into 11 claim families and ranked them from most recent to oldest. Each row separates proof status from evidence strength, because a partially proven claim backed by a teleoperated demo is a very different thing from a claim that was actually fulfilled autonomously in public.
If you want the broader market context on how Tesla Optimus fits within the humanoid robotics sector, we go deeper in our humanoid robotics market deck.
| Capability / Promise | Statement date | Source type | Context | Timeline given? | Has it been proven? | Evidence strength | How it has been proven | Environment of proof | Source |
|---|---|---|---|---|---|---|---|---|---|
| 2025 build target around 5k to 10k units | Mar 2025 | Earnings call / all-hands | Musk first described roughly 10,000 as the internal plan. He later reframed 2025 around 5,000, with 10,000 to 12,000 aspirational. | Yes: 2025 | No | Weak evidence | Public estimates put 2025 output at only a few hundred units. | Controlled internal setting | The Verge |
| Several thousand useful factory bots by end 2025 | Jul 2024 | Earnings call | Musk said Optimus was already doing factory tasks. He forecast several thousand doing useful work by end-2025. | Yes: end of 2025 | No | Weak evidence | Public reporting put 2025 shipments below 500, with no useful-factory proof. | Controlled internal setting | Nasdaq |
| Useful internal Tesla robot use in 2025 | Jul 2024 | X post | Musk said Tesla would have genuinely useful robots next year. He limited first use to Tesla internally. | Yes: 2025 | Partially | Weak evidence | Robots appear to exist internally, but useful deployed work remains unproven publicly. | Controlled internal setting | Yahoo Tech |
| Factory tasks performed by end of 2024 | Apr 2024 | Earnings call | Musk said Optimus could do factory tasks this year. He gave investors a near-term factory milestone. | Yes: end of 2024 | No | Weak evidence | Only lab-like demos surfaced publicly; no production-factory proof by 2024. | Lab demo | Inc. (Reuters) |
| Ready for sale by end of 2025 | Apr 2024 | Earnings call | Musk said it could be ready for sale soon. He put that as early as end-2025. | Yes: end of 2025 | No | No evidence | Optimus was not publicly on sale by end-2025. | Unknown | Inc. (Reuters) |
| Use tools with dexterous production hands | Sep 2022 | Stage presentation | Musk said production hands would move fingers independently. He said that should enable tools and useful work. | No precise timeline | Partially | Weak evidence | Tesla showed manipulation demos, but notable later hand demos were teleoperated. | Teleoperated setting | YouTube (AI Day 2022) |
| Unit cost under $20,000 at scale | Sep 2022 | Stage presentation | Musk said Optimus should cost much less than a car. He guessed below $20,000. | No precise date | No | No evidence | No commercial product or verified sale price exists. | Unknown | YouTube (AI Day 2022) |
| Prototype shown the following year | Aug 2021 | Stage presentation | Musk said Tesla should have a prototype next year. He said the bot would be real. | Yes: sometime in 2022 | Yes | Strong evidence | A working prototype appeared publicly at AI Day 2022. | Stage demo | YouTube (AI Day 2021) |
| Attach bolts to a car with tools | Aug 2021 | Stage presentation | Musk described bolting a car part with a wrench. He framed it as a spoken-command task. | No precise timeline | No | No evidence | No credible public proof of autonomous bolting on a car line. | Unknown | YouTube (AI Day 2021) |
| Fetch groceries from a store | Aug 2021 | Stage presentation | Musk said it should fetch groceries from a store. He used it as a household example. | No precise timeline | No | No evidence | No credible public evidence of real shopping-task completion. | Unknown | YouTube (AI Day 2021) |
| Navigate the human world safely | Aug 2021 | Stage presentation | Musk said it would navigate a world built for humans. He also stressed friendliness and safety. | No precise timeline | Partially | Weak evidence | Public demos show walking and simple movement, not robust real-world autonomy. | Controlled internal setting | YouTube (AI Day 2021) |
This market map, featured in our humanoid robotics market deck, highlights top companies and startups in the humanoid robotics market
How many Optimus promises has Tesla fulfilled so far?
Out of the 11 Optimus promises we tracked from Elon Musk, only one has clearly been fulfilled: the August 2021 commitment that Tesla would show a working prototype the following year.
That 2021 promise was specific enough to score cleanly. Musk said Tesla would show a prototype "sometime next year," and a working prototype did appear on stage at AI Day 2022. Both the capability and the timeline lined up.
Every other Optimus promise is still hanging. That includes factory tasks by end-2024, retail sale by end-2025, the sub-$20,000 unit cost, autonomous tool use, grocery runs, and every single 2025 production target Musk has given publicly.
Some readers might count the fact that a handful of Optimus robots now exist inside Tesla facilities as partial progress on the "useful internal use in 2025" claim. We don't count that as fulfilled, because Musk's exact word was "genuinely useful," and no one outside Tesla has seen evidence of the robots doing useful, paid work.
By the way, if you want to see how Tesla's fulfilled-promise record compares to Agility, Figure, Boston Dynamics, Unitree, and the rest of the humanoid sector, we go deeper in our humanoid robotics market deck.
How many Optimus claims remain unproven today?
Ten out of the 11 Optimus promises we tracked from Musk are still unproven as of April 2026, including every capability claim more ambitious than showing a prototype.
Only the 2022 prototype reveal is publicly validated. A more generous reading, which credits the fact that Tesla has some Optimus units inside its own facilities, would leave nine claims unproven instead of ten.
The weakest evidence clusters around three themes: household tasks like fetching groceries, the promised sub-$20,000 unit cost, and the 2025 production volume targets. In all three, there is nothing credible in public view that matches the original claim.
Everything Tesla has publicly shown stays at the level of "a robot exists and can walk or move objects." Nothing in the public record yet proves "a robot is doing paid, useful work at scale."
How many Optimus milestones missed their original deadlines?
Four to five Optimus milestones have now missed Musk's stated deadlines, all by at least four months.
The oldest and most specific miss is the April 2024 promise that Optimus would handle factory tasks by the end of 2024. As of April 2026, that one is already more than 15 months past due, with no public evidence that the capability ever landed in production.
Here are the tracked deadline misses and how overdue each one is today:
- Factory tasks by end of 2024: missed by about 15 to 16 months with no production proof
- Ready for sale by end of 2025: missed by about 4 months with no public sale
- Useful internal Tesla use in 2025: missed by about 4 months under a strict reading
- Several thousand useful factory bots by end of 2025: missed by about 4 months
- 2025 build target of 5,000 to 10,000 units: missed by about 4 months, with actual output far below even the reduced target
This chart, featured in our humanoid robotics market deck, compares the main business model options for humanoid robot manufacturers
How many Tesla Optimus units are operating inside factories today?
As far as the public record shows, zero Optimus units are doing economically productive factory work today.
Tesla has not disclosed a single uptime, cycle time, throughput, or labor-substitution number for any Optimus robot working on a production line. In a program actually running at scale, at least one of those numbers would normally be public.
It is possible that a few dozen test units physically exist inside Tesla facilities. We would put that broader "physical presence" count somewhere between zero and 100 units, though nothing in the public record confirms even that range.
Tesla's own January 2026 quarterly deck still refers to Gen 3 as the first Optimus design built for mass production, with start of production positioned before end of 2026. That is the language of a program getting ready to ramp, not one that is already running.
By the way, if you want to see how Optimus factory-presence numbers compare to Agility, Figure, and the other humanoid programs, we go deeper in our humanoid robotics market deck.
How many factory tasks does Optimus perform in production today?
No Optimus factory task has been publicly verified as running in real production today, only controlled or staged demos.
Public Optimus demos have shown walking, object transport, and hand manipulation. Every one of them happened inside a stage, a lab, or a controlled Tesla setting, not on a live manufacturing flow where reliability and throughput can be measured.
Several of the most impressive 2024 Optimus demos turned out to be remotely operated by humans. Bloomberg reported this about the October 2024 Cybercab event. That one revelation undermines a lot of the 2024 Optimus communications layer, because if the most visible demos were teleoperated, it gets harder to trust the rest.
The distinction between a demo task and a production task matters here. A production task means the robot repeats the same autonomous job, reliably, inside a live manufacturing flow. Optimus has not publicly crossed that line yet.
How many times has Musk changed Optimus production targets?
Musk has changed his 2025 Optimus production target at least two to three times in under a year, and the number keeps moving downward.
This matters because the revisions happened in fast succession, not spread out across several years. That kind of target instability is unusual for a program supposedly entering production imminently.
Here is the sequence of 2025 Optimus production targets Musk has given publicly:
- July 2024: "several thousand" useful factory bots by end of 2025
- January 2025: "roughly 10,000" as the internal 2025 production plan
- March 2025: about 5,000 as baseline, with 10,000 to 12,000 framed as aspirational
By the way, if you want to see how these shifting Optimus targets line up against the rest of the humanoid industry's production plans, we go deeper in our humanoid robotics market deck.
This chart, featured in our humanoid robotics market deck, shows how Agility Robotics is capturing share in humanoid robotics
How many Optimus use cases have measurable productivity data?
Zero Optimus use cases have any publicly available productivity data attached to them today.
There is no public cycle time, no uptime figure, no labor-displacement number, no ROI, and no attributable cost savings. Tesla has not published a single operating metric for any deployed Optimus unit.
Productivity data is what distinguishes a working program from a prototype program. Throughput, reliability, cost savings, time saved, units per hour. These are the numbers that would normally anchor an investor narrative about a robot already doing useful work.
The fact that none of these numbers exist publicly is itself the clearest signal. A program actually delivering value usually shows at least one operating figure. Optimus has shown none.
Insights
Across five years of Elon Musk statements, AI Day presentations, earnings calls, and follow-up reporting on Tesla Optimus, some patterns only become visible when claims are placed next to each other. Here is what stood out.
- Out of the 11 Optimus promises we tracked from Musk between 2021 and 2025, only one has been publicly fulfilled. That one is the August 2021 promise to show a working prototype in 2022. Every more ambitious claim about autonomy, usefulness, or scale remains open or missed.
- Tesla's 2022 Optimus prototype reveal happened before Musk started attaching commercial promises to the program. Once commercialization entered the language around 2024, his hit rate on claims dropped sharply. Showing a prototype exists is a much lower bar than proving it is useful or produced at scale.
- Musk's earliest 2021 Optimus claims jumped straight to end-user tasks like bolting car parts and fetching groceries. Tesla had not yet publicly shown the robot walking at that point. The rhetoric skipped past basic robotics milestones and landed directly on consumer-grade usefulness.
- The Optimus statements we tracked lean heavily on task descriptions like "get groceries" rather than on measurable sub-capabilities like grasp success rate or cycle time. That framing makes the promises vivid but hard to verify. It also makes future underperformance easier to reinterpret after the fact.
- Musk's Optimus language describes what the robot will eventually do far more often than what it already does. Across five years of public statements, specific operating numbers are almost absent. The gap between vision language and operations language is structural, not accidental.
- The word "useful" is doing a lot of work across Musk's Optimus claims. Once you enforce it strictly, several near-term milestones fail to qualify. The 2025 internal-use promise is the clearest example: robots exist inside Tesla, but no one has shown them doing useful paid work.
- The proof gap is widest on household Optimus claims. Factory-style demos at least exist in some form. For household tasks like grocery runs, there is still no credible public demonstration at all, five years after Musk first described them.
- Optimus hand dexterity demos got visually more impressive between 2022 and 2024. But several of the most striking 2024 hand demos were later reported as teleoperated by humans. What looked like a steady capability trend turned out to be mostly a presentation trend.
- Bloomberg's October 2024 revelation that some Optimus demos at the Cybercab event were remotely operated does more than weaken that single event. It forces a stricter reading of the entire 2024 Optimus communications layer, including any "personal assistant" or dexterity footage from that period.
- Musk's Optimus factory narrative escalated fast. In April 2024 he said Optimus could do factory tasks that year. By July 2024 he was forecasting several thousand robots doing useful factory work by end-2025. Public proof never caught up to even the earlier, smaller milestone.
- The sharpest Optimus deadline miss is not a grand volume target. It is the specific April 2024 claim that the robot would perform factory tasks by the end of 2024. More than 15 months past that deadline, there is still no public proof it ever happened.
- Musk's July 2024 X post was already walking back his April 2024 tone on Optimus. He moved from "factory capability this year" to "useful humanoid robots in low production for Tesla internal use next year." That shift happened in about three months, before the original deadline even arrived.
- Musk's 2025 Optimus production targets changed materially in a matter of weeks. In January 2025 he described "roughly 10,000" as the internal plan. By March 2025 the wording had softened to about 5,000 baseline. That kind of target instability is unusual for a program supposedly entering production.
- 2024 was the year Musk turned Optimus from robotics talk into concrete commercial timing promises. It is also the year the claims aged worst. The commercialization language is what broke first, before any of the capability language.
- A consistent three-step pattern shows up across the Optimus statements we tracked. First a capability claim, then a production forecast, then public proof, often months or years late, and often absent entirely. Step three has consistently lagged the other two.
- Musk's under-$20,000 cost target for Optimus is one of the least supported claims in the entire record. There is no public bill-of-materials anchor, no launch price, no early commercial pricing band. It is effectively still a stage-presentation number from 2022.
- Public estimates suggest Tesla built fewer than 500 Optimus units in 2025. That number misses every single 2025 target Musk discussed publicly: "several thousand," "roughly 10,000," and "about 5,000." No reading of the record rescues those numbers.
- Every favorable Optimus proof point we found sits in a stage demo, a lab setting, a teleoperated setup, or a controlled Tesla facility. Nothing has been validated in a real customer or live production environment. The gap between "shown" and "deployed" is still the defining gap.
- Optimus proof points cluster into four environments: stage demo, lab demo, teleoperation, and controlled internal use. None of those qualify as "factory production" under a serious industrial standard. The gap is structural and hasn't narrowed meaningfully.
- Tesla's own January 2026 Q4 deck still describes Gen 3 as the first Optimus design built for mass production, with start of production positioned before end of 2026. That means Tesla itself is still treating Optimus as a pre-ramp program, not as an already-useful internal fleet.
- Public evidence suggests Tesla spent 2025 in a design and line-installation phase for Optimus, not in an economically productive deployment phase. Tesla's own terms like "production-primed" and "preparations underway" fit a pre-ramp program, not one already generating useful internal work.
- Tesla talks openly about Optimus annual capacity ambitions but has not published a single operating metric. The willingness to share forward-looking numbers sits alongside a refusal to share performance numbers. That asymmetry is itself a signal worth weighting heavily.
- The overall pattern across five years of Optimus statements is not "Musk said a lot and some came true." It is closer to "Musk proved concept existence early, then consistently overreached on usefulness, timing, scale, and price." Optimus today is still mostly a promise stack, not a deployed product stack.
This chart, featured in our humanoid robotics market deck, illustrates yearly funding for humanoid robotics startups
Our methodology to build this tracker
We limited the corpus to direct first-hand Musk statements only. Those include Tesla AI Day remarks, earnings calls, shareholder-stage remarks, and Musk's own X posts where the wording was reproducible from high-quality reporting or transcripts. When the original source page was easy to identify, we used it. When the original X post or webcast transcript was not directly surfaced, we used high-quality transcript or Reuters-syndicated reproductions that faithfully quote Musk's words.
We grouped repeated statements into claim families. For example, Musk repeated the broad "useful factory robots in 2025" theme multiple times across 2024 and 2025. We treated that as one underlying family unless the wording materially changed the scope, quantity, or deadline. For the 2025 production-target family, the wording changed enough to count as target revisions rather than mere repetitions.
We applied a conservative proof standard. "Yes" means credible public evidence shows the capability in conditions reasonably close to the claim. "Partially" means only a subset was shown, or it was shown in a staged, simplified, teleoperated, or otherwise limited setting. "No" means there is no credible public evidence matching the claim. We scored evidence strength separately. "Strong evidence" means the public proof is direct and fairly close to the claim. "Weak evidence" means suggestive but incomplete or compromised. "No evidence" means there is no meaningful public proof.
We also tracked the environment of proof so stage demos, lab demos, teleoperation, controlled internal tests, and genuine production use would not be conflated. Where exact figures were unavailable, we used bounded estimate ranges instead of false precision. That matters especially for factory deployment counts, production totals, and deadline slippage, because Tesla has disclosed very little hard public operating data for Optimus.
Related blog posts
- How close is Tesla Optimus to being deployed at scale?
- Can we believe the impressive humanoid robot demos?
- The full range of business models in the humanoid robotics market
- How close is Agility Digit to large-scale deployment?
