Signals Inbox·July 15, 2026·Humanoid Robotics

Figure vs Optimus: who is ahead?

Figure is ahead of Tesla Optimus today because it has put more robots into real workplaces, disclosed more operating data and started an actual production ramp. Tesla remains the more dangerous long-term competitor, but most of its advantage still depends on what happens next.

Q1Who is closer to putting a useful humanoid to work, Figure or Tesla Optimus?

Figure is clearly ahead of Tesla Optimus in publicly documented workplace use today.

The strongest evidence comes from BMW’s Spartanburg factory. Figure says its Figure 02 robots accumulated more than 1,250 operating hours, loaded over 90,000 sheet-metal parts and contributed to the production process for more than 30,000 BMW X3 vehicles.

BMW also discussed the pilot publicly, which gives the deployment more weight than an internal company claim. It took place inside a real automotive plant, around existing workers and equipment, with production schedules that Figure could not fully control.

The 90,000-part figure needs some context. Moving one type of component repeatedly is still a narrow task. It does not show that Figure 02 could move freely between ten unrelated factory jobs. It does show that the robot could survive a meaningful amount of repetitive industrial work.

Figure returned to the same BMW facility with Figure 03 in June 2026. That second deployment is arguably more informative than the first announcement. Companies run experiments all the time. A customer bringing the next robot generation back into the factory suggests that the original pilot produced enough value to continue.

Tesla says Optimus has also worked inside its factories. Yet it has released no comparable figures for hours, completed cycles, parts moved or production output. We also lack confirmation from an independent customer because Tesla itself owns the facilities.

Tesla’s internal access still gives it a serious advantage. It can place robots into manufacturing environments without waiting for another company to approve the pilot. Engineers can watch failures directly and change the robot or the workstation quickly.

Even with that advantage, the public comparison is currently one-sided. Figure has given us a named factory, a defined task, operating hours, task volume and a follow-up deployment. Tesla has mostly given us demonstrations and broad statements about internal testing.

Figure is at least one step further along in proving that its robot can contribute to real industrial work.

Q2Who is building the smarter humanoid today, Figure or Tesla Optimus?

Figure is currently ahead of Tesla Optimus in demonstrated humanoid autonomy.

Figure has shown more than isolated manipulation tricks. Its Helix system connects visual understanding, language instructions and continuous control of the upper body, including the wrists, hands, torso and fingers.

The company later expanded that approach through Helix 02, which adds whole-body movement. Recent demonstrations have included opening doors, moving furniture, picking up varied objects, cleaning rooms and coordinating actions across longer sequences.

Those sequences tell us more than a robot placing the same object into the same container. The system has to remember what has already happened, move through the room, reposition its body and choose the next action.

Figure has also demonstrated two robots cooperating on a shared task. Multi-robot coordination remains early, but it starts to show how a future fleet could divide work rather than operate as isolated machines.

Tesla has released videos of Optimus walking, sorting objects, carrying components and completing household-style tasks. The company also has deep expertise in computer vision, neural-network training and real-time decision systems through its vehicle program.

The problem is that Tesla explains much less about what happens inside the robot. It rarely discloses the number of attempts, the amount of human assistance or how the system responds when an object moves unexpectedly.

Some previous Optimus demonstrations involved teleoperation. Teleoperation still creates useful training data, and nearly every humanoid company uses it. It becomes misleading only when viewers cannot tell whether the robot is acting autonomously.

Figure’s demonstrations are company-controlled too. We do not see complete failure logs or a full account of interventions. Figure has not proved that Helix can run independently for a full working shift.

Still, the gap is fairly clear today. Figure has shown broader sequences and published a more coherent description of its humanoid AI architecture. Tesla may have powerful internal systems, but it has supplied too little evidence to receive equal credit.

Q3Who is ahead in reliable factory work, Figure or Tesla Optimus?

Figure is ahead of Tesla Optimus in demonstrated factory reliability, although neither company has published enough data to prove full commercial readiness.

Figure’s BMW figures show repeated operation over a meaningful period. The company says Figure 02 worked ten-hour weekday schedules, travelled the equivalent of more than 200 miles and handled more than 90,000 parts.

That is a stronger test than a short factory video. Repetitive loading exposes weaknesses in joints, hands, balance, batteries and software. Small errors that look harmless in a demonstration become expensive when they happen hundreds of times per shift.

The missing figures are just as important. Figure has not disclosed uptime, intervention frequency, maintenance hours or how often the robot failed to complete a cycle. It has also said little about the number of robots involved, which prevents us from calculating useful output per robot.

Without those details, we cannot tell whether Figure 02 worked for most of each ten-hour period or spent significant time waiting for assistance.

Tesla has released even less. There are no recent public Optimus statistics that can be compared directly with Figure’s factory hours. Tesla may have extensive internal logs, but those logs remain private.

This gives Figure a genuine lead with a wide confidence range. The BMW deployment proves that Figure can operate in a factory repeatedly. It does not prove that the robot matches the availability of conventional automation or the economics of a human worker.

The next useful disclosure from either company should include interventions per shift, successful cycles, downtime, maintenance and cost per productive hour. Another polished video would add very little.

Q4Who has the more advanced humanoid hardware today, Figure or Tesla Optimus?

Figure is slightly ahead of Tesla Optimus in hardware that has actually reached production, while Tesla’s next hand could change the comparison quickly.

Figure designed Figure 03 around more than performance. The company says it replaced several expensive manufacturing processes with die casting, stamping and injection molding. It also brought more battery development and production inside the company.

Figure is already dealing with the unglamorous problems that separate a product from a laboratory prototype: assembly time, component cost, thermal management, serviceability and production yield.

Figure 03 also includes improved tactile sensing and a redesigned hand. The hand has to manage objects with different shapes, materials and levels of fragility. It also needs to survive repeated impacts and thousands of grasping cycles.

Tesla has repeatedly described the Optimus hand as one of its hardest engineering challenges. Elon Musk has said the next version uses a substantially redesigned system with more sophisticated actuation.

A major improvement in hand dexterity could move Tesla ahead. Most useful factory and household tasks demand much more from the hands than from the legs. Walking through a room has little value when the robot cannot grip a thin component, rotate a tool or control force precisely.

Today, however, Tesla’s latest design remains largely hidden. The company says it wants to limit copying before the production version is ready. Fair enough, but it leaves outsiders unable to judge the hardware.

Figure has already built hundreds of Figure 03 units. Production volume alone does not prove that the hand is superior, but it shows that the hardware design has moved further through assembly, testing and manufacturing.

Figure leads in what has been built and disclosed. Tesla’s potential lead depends on hardware that still needs to be shown, produced and tested over thousands of cycles.

Q5Who is learning faster, Figure or Tesla Optimus?

Neither Figure nor Tesla Optimus has proved a clear lead in learning speed. Each company has a different advantage.

Tesla may have access to the richer industrial environment. Its robots can train inside car factories filled with repeatable tasks, human demonstrations, production equipment and measurable outcomes.

The Optimus team can also use Tesla’s wider AI infrastructure. The company already trains large vision models, operates substantial compute systems and employs engineers experienced in turning visual data into real-time actions.

That gives Tesla a strong base, but driving and manipulation are very different problems. A vehicle observes the road and controls a small set of outputs. A humanoid hand has many joints and must handle objects that bend, slip, rotate and disappear behind the fingers.

Figure has explained its humanoid learning strategy in more detail. Helix uses robot demonstrations, visual inputs, language instructions and human-motion data. The company has also discussed collecting larger volumes of human video and sensory information to pretrain future systems.

This approach could help Figure learn a broader range of human activities without collecting every example directly on a robot. A person can demonstrate folding clothes or arranging objects far more cheaply than a humanoid can repeat the task thousands of times.

Human video has limits. It rarely contains accurate force information, tactile feedback or joint-level robot commands. Figure still needs real robot experience to learn what happens when an object slips or a grasp fails.

Tesla’s internal factories may generate more directly useful industrial data. Figure’s method may generalize better across different environments. Neither company has published the key result: how much faster the robot learns its tenth new task compared with its first.

Until we see that improvement curve, the category remains open. Tesla leads in potential access to industrial data, while Figure gives us a clearer view of how its training pipeline works.

Q6Who is producing more humanoid robots today, Figure or Tesla Optimus?

Figure is clearly producing more publicly documented humanoid robots than Tesla today.

In April 2026, Figure said it had produced more than 350 Figure 03 robots. It also reported moving from roughly one robot per day to one per hour in less than four months.

A production rate of one per hour would equal several thousand robots per year if sustained across normal factory schedules. That remains below Figure’s stated BotQ capacity of 12,000 units annually, but it shows that the company has started a genuine manufacturing ramp.

The 350-unit figure also places Figure ahead of most Western humanoid startups in disclosed production. Agility Robotics has built a dedicated facility and Apptronik is preparing larger commercial deployments, although few US competitors have publicly confirmed hundreds of current-generation units from one line.

Tesla has much larger production ambitions. It has discussed converting former Model S and Model X space in Fremont into an Optimus line and eventually reaching one million robots per year there.

Those targets sit in a completely different league from Figure’s initial factory. Yet Tesla had not disclosed a comparable number of finished, current-generation Optimus robots by mid-July 2026.

The distinction is simple. Figure is already ramping a line. Tesla is preparing a line that could eventually become far larger.

Figure’s current lead could disappear fast. At one robot per hour, it might take Figure a year to produce the number of robots Tesla could theoretically build in a few days at full future capacity. Tesla first has to finish the design, install the equipment and prove that the line works.

Today’s production leader is Figure. The future volume leader is still undecided.

Q7Who is winning more real customers, Figure or Tesla Optimus?

Figure is clearly ahead of Tesla Optimus in external customer validation.

BMW remains Figure’s strongest customer reference. The relationship has continued across two robot generations, beginning with Figure 02 and expanding to Figure 03 in June 2026.

That continuation separates the BMW project from many humanoid pilots. A large manufacturer can run a short trial for publicity, learning or strategic exploration. Bringing the next version back suggests that the customer sees enough operational value to keep investing time in the relationship.

Figure also signed an agreement with Catalyst Brands for an initial logistics deployment in Reno. The planned tasks focus on physically demanding distribution work.

This second customer broadens the evidence. Automotive part loading and apparel distribution involve different workflows, objects and operating conditions. Success in both would strengthen Figure’s claim that its platform can travel across industries.

The commercial details remain thin. Figure has not disclosed how many robots each customer will use, how much they are paying or whether the agreements include binding volume commitments.

Figure has proved customer access and continued interest. It has not yet proved a large, repeatable robot business.

Tesla can begin without outside customers. Its factories provide a huge internal market, and Tesla can measure value directly through labor savings, throughput or safety improvements.

Even so, internal adoption cannot answer every commercial question. Tesla can tolerate early inefficiency for strategic reasons. An external buyer has less patience and can walk away when the economics fail.

Figure currently has the stronger market test. BMW returned for another deployment, and Catalyst Brands gives Figure a second industrial setting. Tesla has yet to show that an independent company is ready to deploy Optimus.

Q8Who could manufacture humanoid robots at the largest scale, Figure or Tesla Optimus?

Tesla is far ahead of Figure in potential manufacturing scale, even though Figure has executed more of the humanoid production ramp so far.

Tesla already manufactures complex products in enormous volumes. It controls large factories, global supplier relationships, battery expertise, electronics production and advanced automation.

In the second quarter of 2026, Tesla produced more than 450,000 vehicles. A humanoid robot differs greatly from a car, but Tesla already knows how to manage thousands of components, quality systems and high-cost production equipment across multiple factories.

Figure is building those capabilities almost from scratch. BotQ is designed specifically for humanoid robots, which gives the company focus. Its first planned annual capacity of 12,000 units remains small beside Tesla’s proposed one-million-unit Fremont line.

Tesla can also invest much more aggressively. It could order tooling, secure components and tolerate years of losses in a way that would be difficult for most startups.

Still, Tesla’s scale advantage begins only after Optimus reaches a stable production design. A robot with unreliable hands or short component life cannot be rescued by a larger factory. Producing defects faster would simply increase losses.

Figure has already worked through an initial version of that transition. It has adapted Figure 03 for manufacturing, started production and increased line speed. Its advantage lies in current execution rather than ultimate capacity.

Tesla’s ceiling is much higher. If Optimus proves dependable and the Fremont line starts running effectively, Tesla could pass Figure’s cumulative production within a short period.

As things stand in July 2026, Figure knows more about producing its current humanoid. Tesla knows far more about building complex machines at industrial scale. The second advantage will eventually carry more weight, provided Optimus becomes ready for it.

Q9Who gives us stronger evidence of progress, Figure or Tesla Optimus?

Figure gives us far stronger evidence than Tesla about where its humanoid program stands today.

Figure publishes named customers, deployment dates, robot counts, operating hours and technical explanations of Helix. It has also described changes to the battery, hands and production process.

The disclosures are still selective. Figure controls its videos and chooses which metrics become public. We rarely see failures, interrupted runs or complete economics.

Yet the available details allow a serious comparison. We can ask whether 1,250 hours is enough, whether 350 robots is a meaningful production run and whether BMW’s second deployment strengthens the first.

Tesla provides fewer points of reference. Shareholder updates discuss Optimus, future factories and expected production, but they rarely include current robot counts, runtime or task success rates.

This makes almost every Tesla comparison more uncertain. A lack of disclosure does not show that Optimus is failing. It means we cannot separate internal progress from ambitious statements.

Tesla may have good competitive reasons for secrecy. Humanoid designs are evolving quickly, and hand mechanisms can be copied. Still, secrecy shifts the burden onto Tesla. The company cannot claim the credibility of a proven deployment while keeping the evidence private.

Figure’s transparency creates a practical competitive advantage in the race for customers, employees and partners. Companies considering a pilot can examine a more concrete record.

For readers trying to judge the race today, Figure gives us much firmer ground. Tesla asks us to trust that its industrial capabilities will eventually translate into a strong humanoid.

Q10Who is ahead overall today, Figure or Tesla Optimus?

Figure is ahead of Tesla Optimus today, although Tesla remains the more dangerous long-term competitor.

Figure currently wins the categories that tell us whether a humanoid is becoming useful. It has stronger external deployment evidence, broader autonomy demonstrations, more disclosed factory runtime, hundreds of produced robots and two named customer relationships.

Those leads are connected. Figure is building robots, placing them in workplaces, gathering data and feeding what it learns into the next hardware and software versions. That looks like the beginning of a commercial loop.

The loop is still small. BMW has not announced a large fleet order. Catalyst Brands is beginning with an initial deployment. Figure has not published the cost per robot-hour or enough reliability data to show that its machines can compete economically with people and conventional automation.

Tesla’s strengths sit further downstream. It has larger factories, more capital, stronger supply-chain leverage and a much bigger internal deployment environment. Once Optimus works reliably, Tesla could manufacture and deploy it at a pace Figure would struggle to match.

Today, the phrase “once Optimus works reliably” carries most of the uncertainty. Tesla has not published enough evidence about its latest robot, its autonomy or its factory performance.

Figure’s overall lead is best described as one stage of commercial development. It has reached externally validated pilots and early production. Tesla is preparing for a much larger production phase without publicly proving that the robot is ready for it.

Figure could extend its lead during the rest of 2026 by increasing its BMW fleet, completing the Catalyst deployment and publishing intervention, uptime and economics data. Those results would move the company beyond promising pilots.

Tesla could erase the gap by unveiling its new Optimus generation, producing it in the thousands and showing sustained autonomous work inside its factories. A public dataset covering hours, output and interventions would change this comparison more than another product demonstration.

As of today, Figure is closer to having a commercially useful humanoid. Tesla remains better placed to dominate manufacturing later, but it still has to prove that Optimus is ready to be manufactured at that scale.

Figure vs Tesla Optimus, July 2026