WEBVTT
00:00:00.000 --> 00:00:02.720
Tesla Optimus vs Aptronic Apollo.
00:00:02.959 --> 00:00:06.160
Which 2026 robot is closer to factory scale deployment?
00:00:06.320 --> 00:00:09.359
By Akari Sato, consumer robotics specialist.
00:00:09.599 --> 00:00:12.720
Status checked through October 29, 2026.
00:00:12.880 --> 00:00:16.000
Tesla is closer to building humanoid robots at scale.
00:00:16.239 --> 00:00:19.920
Aptronic is closer to testing them with outside industrial partners.
00:00:20.079 --> 00:00:27.600
Neither has shown that its latest robot can do useful factory work at scale, for a full shift, at a proven return on investment.
00:00:27.839 --> 00:00:29.199
That difference matters.
00:00:29.440 --> 00:00:35.439
A factory that makes robots is not the same thing as a factory where robots reliably make products.
00:00:35.759 --> 00:00:39.039
Tesla has started early production of Optimus in Fremont.
00:00:39.280 --> 00:00:48.399
Apptronic has built Apollo 2 robots for training and pilots, and has named partners including Mercedes Benz, GXO, and Jabble.
00:00:48.560 --> 00:00:55.119
But neither company has published the operating numbers a buyer needs to confirm dependable, economical factory work.
00:00:55.280 --> 00:00:58.479
The short answer, the winner depends on what closer means.
00:00:58.640 --> 00:01:01.840
Closer to high-volume robot manufacturing, Tesla.
00:01:02.000 --> 00:01:07.680
Tesla says its Fremont line has begun producing its first production quality Optimus robots.
00:01:07.920 --> 00:01:13.120
It has discussed a long-term capacity target of 1 million robots a year for that line.
00:01:13.280 --> 00:01:20.079
But the latest shareholder update lists Optimus as early ramp and gives no installed capacity figure.
00:01:20.400 --> 00:01:26.480
CEO Elon Musk said it could take at least a year to reach about 20,000 robots per week.
00:01:26.640 --> 00:01:29.200
That is a target, not current output.
00:01:29.359 --> 00:01:32.560
Closer to named third-party trials, Aptronic.
00:01:32.879 --> 00:01:36.400
Apollo 2 is being used for data collection and pilot work.
00:01:36.480 --> 00:01:40.719
And Aptronic says its robots are active at customer and partner sites.
00:01:41.040 --> 00:01:49.120
Reuters reported that Aptronic had built hundreds of Apollo 2 robots by June, but the company did not disclose how many were deployed.
00:01:49.359 --> 00:01:52.400
Closer to proven production economics, neither.
00:01:52.640 --> 00:01:56.480
Aptronic's named relationships are not evidence of scaled production.
00:01:56.640 --> 00:02:09.840
For context, logistics partner GXO said it had completed 45 humanoid pilots across the industry, had not reached return on investment with humanoids, and expected zero humanoids in production at GXO in 2026.
00:02:10.159 --> 00:02:15.599
That figure is not specific to Apollo, but it's a useful reality check for the whole sector.
00:02:15.919 --> 00:02:19.520
My verdict, Tesla leads on the factory that builds the robot.
00:02:19.759 --> 00:02:22.960
Aptronic leads on visible customer and partner engagement.
00:02:23.199 --> 00:02:26.800
Neither has yet proved the robot as a dependable production worker.
00:02:26.960 --> 00:02:28.800
What versions are we comparing?
00:02:29.039 --> 00:02:40.560
As of October 29, 2026, Tesla's latest production effort is associated with Optimus Gen 3, which Tesla has described as its first design intended for mass production.
00:02:40.800 --> 00:02:44.159
But Tesla has not published a complete current spec sheet for it.
00:02:44.240 --> 00:02:51.039
The company's October update confirms early production and says the manufacturing process is suitable for a commercial product.
00:02:51.280 --> 00:02:57.360
It does not disclose the robot's detailed dimensions, payload, runtime, or current task results.
00:02:57.599 --> 00:03:05.759
Aptronic's latest disclosed platform is Apollo 2, unveiled on June 30, 2026, in both bipedal and wheeled versions.
00:03:06.000 --> 00:03:10.080
Aptronic describes it as a data collection and training platform for pilots.
00:03:10.240 --> 00:03:17.680
The company calls Apollo 3 its next generation commercial product, but has not published its specifications or affirm launch date.
00:03:17.919 --> 00:03:28.240
CEO Jeff Cardenas told Reuters that pilot work would continue through 2026, with real production versions expected in 2027 and beyond.
00:03:28.479 --> 00:03:35.680
A major caution: Aptronic's well-known numbers for height, weight, lift, and battery runtime were published for the original Apollo.
00:03:35.840 --> 00:03:39.599
They should not be treated as confirmed Apollo 2 specifications.
00:03:39.840 --> 00:03:43.360
The same caution applies to Tesla's early concept measurements.
00:03:43.520 --> 00:03:45.360
They are not Gen 3 measurements.
00:03:45.599 --> 00:03:51.439
Hardware comparison.com.
00:03:51.680 --> 00:03:57.199
The table shows a real difference in disclosure, not necessarily a difference in hardware quality.
00:03:57.439 --> 00:04:01.199
Apptronic publishes more about battery swapping and safety zones.
00:04:01.360 --> 00:04:04.960
Tesla publishes more about factory investment and production plans.
00:04:05.120 --> 00:04:10.319
Neither gives buyers the operating data needed for a full cost and reliability comparison.
00:04:10.639 --> 00:04:12.319
Autonomy and perception.
00:04:14.879 --> 00:04:20.480
Tesla's pitch is that its experience in vehicle AI will help optimists learn real-world tasks.
00:04:20.639 --> 00:04:23.040
But factory manipulation is not driving.
00:04:23.279 --> 00:04:26.800
A robot must grip, lift, place, and recover from small errors.
00:04:26.959 --> 00:04:34.399
Tesla has not published a comparable Gen 3 task success rate, cycle time, or full shift autonomy result in its latest update.
00:04:34.560 --> 00:04:37.600
The Q3 2026 call points to the remaining work.
00:04:37.839 --> 00:04:46.319
Musk said Tesla needs roughly 10,000 to 20,000 robots for an Optimus Academy, where robots can practice tasks and generate learning data.
00:04:46.560 --> 00:04:55.279
He also put outside company deliveries around the end of 2027 and said reaching about 20,000 robots per week at Fremont could take at least a year.
00:04:55.519 --> 00:04:59.600
Those are company forecasts, not confirmed dates or output levels.
00:04:59.920 --> 00:05:02.560
Apptronic has a more visible data pipeline.
00:05:02.800 --> 00:05:10.160
It says Apollo 2 gathers data through both teleoperation, a human steering or guiding the robot, and autonomous execution.
00:05:10.319 --> 00:05:15.439
Apptronic and Google DeepMind intend to use that data to improve future robot models.
00:05:15.680 --> 00:05:22.160
This is a thoughtful plan, but teleoperated data collection is not the same as autonomous production work.
00:05:22.399 --> 00:05:26.399
Apptronic does not publish the share of work done without human help.
00:05:26.639 --> 00:05:31.839
Google DeepMind has published test results for Gemini Robotics 2 on Apollo 2.
00:05:31.920 --> 00:05:43.839
In whole body tests, the robots' reported success rates included 68.4% for picking up from a table, 45.7% from the floor, and 76.3% from a shelf.
00:05:44.079 --> 00:05:49.199
In multi-finger tests using a sharp wave hand, results varied widely by task.
00:05:49.519 --> 00:05:56.720
92% for unscrewing a bulb, but 32% for a dustpan task, and 40% for sealing a bag.
00:05:56.959 --> 00:05:59.920
Google notes that movement speed still needs work.
00:06:00.160 --> 00:06:04.000
These are useful research results, not factory throughput figures.
00:06:04.240 --> 00:06:05.519
Oh, interpretation.
00:06:05.759 --> 00:06:11.120
Aptronic currently has the strenger public evidence for how an AI model handles specific robot tasks.
00:06:11.279 --> 00:06:13.920
The results also show how much remains to be done.
00:06:14.160 --> 00:06:22.240
Tesla's large vehicle data advantage may help with seeing and learning, but it does not remove the hard work of reliable grasping and physical task execution.
00:06:22.560 --> 00:06:23.680
Factory pilots.
00:06:23.839 --> 00:06:26.160
Partners are not the same as production customers.
00:06:26.399 --> 00:06:28.480
Tesla a robot production line.
00:06:28.720 --> 00:06:31.120
But limited public proof of robot work.
00:06:31.279 --> 00:06:36.480
Tesla's official materials show a long shift from pilot plans to robot manufacturing.
00:06:36.720 --> 00:06:43.279
In 2024, its investor update said Optimus was performing its first factory task, handling batteries.
00:06:43.439 --> 00:06:51.279
But by Tesla's Q4 2025 earnings call, Musk said Optimus was still in RD and that factory use was not material.
00:06:51.439 --> 00:06:58.959
Now Tesla says Optimus production has begun on Fremont's high volume lines and calls the units the first production quality bots.
00:06:59.120 --> 00:07:01.600
That is meaningful progress in making the robot.
00:07:01.759 --> 00:07:10.399
But the update gives no count of robots deployed on production tasks, no task cycle times, no reliability figures, and no productivity gains.
00:07:10.639 --> 00:07:16.079
The company's latest official table lists Fremont as early ramp and Texas as construction.
00:07:16.560 --> 00:07:18.720
Tesla's manufacturing targets are large.
00:07:18.800 --> 00:07:26.160
It has described a Fremont line designed for 1 million units a year, and a future Texas line designed for up to 10 million a year.
00:07:26.399 --> 00:07:29.199
But those are capacity goals, not current production.
00:07:29.360 --> 00:07:37.600
Tesla's own Q3 table lists no installed capacity number for Optimus, and Musk says the Fremont line still needs time to ramp.
00:07:37.839 --> 00:07:41.759
Apptronic, more named partners, but still pilot stage evidence.
00:07:41.920 --> 00:07:44.399
Apptronic has named several industrial partners.
00:07:44.560 --> 00:07:48.639
Mercedes-Benz, a commercial agreement to pilot Apollo in manufacturing.
00:07:48.800 --> 00:07:53.759
The proposed work includes bringing parts or kits to production workers and checking components.
00:07:54.000 --> 00:07:58.240
The announcement describes exploration and piloting, not a scaled fleet.
00:07:58.560 --> 00:08:03.199
GXO Logistics, an early proof of concept and warehouse testing relationship.
00:08:03.439 --> 00:08:08.480
GXO describes Apollo as one prototype among several humanoids it is evaluating.
00:08:08.800 --> 00:08:11.680
Jable, a pilot and manufacturing partnership.
00:08:11.839 --> 00:08:19.279
The announced tasks include inspection, sorting, kiting, line side delivery, fixture placement, and subassembly.
00:08:19.519 --> 00:08:22.720
Jable is also intended to help manufacture Apollo.
00:08:22.879 --> 00:08:26.959
The public announcement does not report output, uptime, or savings.
00:08:27.199 --> 00:08:31.040
Okay, Google DeepMind, an AI research partner, not a factory buyer.
00:08:31.199 --> 00:08:35.679
Apptronics says Apollo 2 data is being used to develop robotics AI models.
00:08:35.919 --> 00:08:42.080
Reuters reported that Apptronic had built hundreds of Apollo 2 robots but did not disclose deployment counts.
00:08:42.320 --> 00:08:48.480
That is evidence of hardware building experience, not proof that hundreds are working independently in factories.
00:08:48.720 --> 00:08:52.000
Apptronic says Apollo 2 is a training and data platform.
00:08:52.159 --> 00:08:54.159
The commercial product is expected later.
00:08:54.399 --> 00:08:55.279
Practical reading.
00:08:55.440 --> 00:08:58.159
Apptronic has a stronger map of named partner trials.
00:08:58.320 --> 00:09:01.440
Tesla has a stronger visible investment in making robots.
00:09:01.600 --> 00:09:06.480
Neither has released a public record of sustained factory output from a fleet of humanoids.
00:09:06.720 --> 00:09:08.320
Price and economic value.
00:09:08.559 --> 00:09:12.320
Neither company offers a public purchase price for the 2026 robot.
00:09:12.559 --> 00:09:20.159
Tesla's often-quoted$20,000 to$30,000 figure is Elon Musk's long-term estimate for Optimus at scale.
00:09:20.320 --> 00:09:24.639
It is not a Gen 3 quote, and Tesla has not disclosed the current cost to build one.
00:09:24.879 --> 00:09:27.759
Aptronic has not published an Apollo 2 price.
00:09:27.919 --> 00:09:35.679
Coverage has reported that B Capital's Howard Morgan expected Apollo orders in 2027 at roughly$80,000 per robot.
00:09:35.840 --> 00:09:40.559
That is an investor's forecast, not an Atronic price list or signed purchase order.
00:09:40.720 --> 00:09:44.000
It should not be used as a firm cost in a business case.
00:09:44.240 --> 00:09:48.559
For a plant buyer, sticker price is only one part of the decision.
00:09:48.799 --> 00:09:52.159
The useful number is cost per successful task.
00:09:52.399 --> 00:10:03.120
Include robot purchase or lease, software, integration, battery packs, replacement hands and actuators, safety changes, service, human supervision, and downtime.
00:10:03.279 --> 00:10:08.799
Then compare that total with the cost and speed of the process the robot would actually handle.
00:10:09.039 --> 00:10:16.240
A robot that is cheap but needs constant help may cost more than a robot with a higher price and better uptime.
00:10:16.399 --> 00:10:21.360
And a purpose-built machine may beat either humanoid if the task never changes.
00:10:21.600 --> 00:10:26.320
What executives and investors should watch next before approving a large deployment?
00:10:26.559 --> 00:10:30.799
Ask for results that neither company has publicly supplied in a complete form.
00:10:30.960 --> 00:10:34.080
How many hours did the robot work without a person guiding it?
00:10:34.240 --> 00:10:39.200
How many task cycles did it complete per hour compared with a human or existing machine?
00:10:39.440 --> 00:10:42.320
How often did it fail, stop, or need a reset?
00:10:42.559 --> 00:10:46.080
How long does it take to replace a battery or repair a hand?
00:10:46.320 --> 00:10:50.480
Can it keep working across a full shift under real factory condition?
00:10:51.360 --> 00:10:54.639
Ons, what is the all-in cost per completed task?
00:10:54.799 --> 00:10:58.480
For a pilot, set a narrow task and a pass-fail bar.
00:10:58.720 --> 00:11:07.039
For example, move a defined part between two known stations at a target rate, with a set limit on human interventions and downtime.
00:11:07.440 --> 00:11:11.519
Do not accept a stage demo as a substitute for a measured production test.
00:11:11.840 --> 00:11:17.360
Final verdict: Tesla Optimus is closer to factory scale manufacturing of humanoid robots.
00:11:17.600 --> 00:11:21.279
Aptronic Apollo is closer to external industrial pilot work.
00:11:21.519 --> 00:11:25.919
Tesla's Fremont Line and production quality units are the clearest scale signal.
00:11:26.159 --> 00:11:35.200
Aptronic's named partners, data collection program, swappable battery plan, and published Google DeepMind tests give buyers more to inspect today.
00:11:35.440 --> 00:11:40.159
But neither robot is ready to be treated as proven factory labor at scale.
00:11:40.399 --> 00:11:44.639
The next milestone is not a bigger factory or a more impressive video.
00:11:44.799 --> 00:11:52.720
It is a customer publishing the numbers, autonomous hours, task speed, failure rate, repair time, and cost per task.
00:11:52.960 --> 00:12:01.120
Until then, Tesla is the stronger scale bet, Haptronic is the stronger pilot bet, and both remain early stage deployment projects.
00:12:01.279 --> 00:12:05.039
All links to sources are available in the text version of this article.
00:12:05.279 --> 00:12:09.360
You can find the full article at robotcomparisons.com slash blog.
00:12:09.600 --> 00:12:10.559
Thanks for listening.
00:12:10.799 --> 00:12:12.960
Thanks for listening to Robot Comparisons.
00:12:13.120 --> 00:12:18.159
If you found this episode valuable, make sure to subscribe so you never miss a new deep dive.
00:12:18.320 --> 00:12:25.279
For the full written breakdown, detailed spec sheets, and side by side comparisons, head over to robotcomparisons.com.
00:12:25.360 --> 00:12:30.559
That's R O B O T C O M P A R I S O N S dot com.
00:12:30.879 --> 00:12:33.120
New research drops multiple times a week.
00:12:33.360 --> 00:12:34.559
See you in the next one.
00:00:00.000 --> 00:00:02.720
Tesla Optimus vs Aptronic Apollo.
00:00:02.959 --> 00:00:06.160
Which 2026 robot is closer to factory scale deployment?
00:00:06.320 --> 00:00:09.359
By Akari Sato, consumer robotics specialist.
00:00:09.599 --> 00:00:12.720
Status checked through October 29, 2026.
00:00:12.880 --> 00:00:16.000
Tesla is closer to building humanoid robots at scale.
00:00:16.239 --> 00:00:19.920
Aptronic is closer to testing them with outside industrial partners.
00:00:20.079 --> 00:00:27.600
Neither has shown that its latest robot can do useful factory work at scale, for a full shift, at a proven return on investment.
00:00:27.839 --> 00:00:29.199
That difference matters.
00:00:29.440 --> 00:00:35.439
A factory that makes robots is not the same thing as a factory where robots reliably make products.
00:00:35.759 --> 00:00:39.039
Tesla has started early production of Optimus in Fremont.
00:00:39.280 --> 00:00:48.399
Apptronic has built Apollo 2 robots for training and pilots, and has named partners including Mercedes Benz, GXO, and Jabble.
00:00:48.560 --> 00:00:55.119
But neither company has published the operating numbers a buyer needs to confirm dependable, economical factory work.
00:00:55.280 --> 00:00:58.479
The short answer, the winner depends on what closer means.
00:00:58.640 --> 00:01:01.840
Closer to high-volume robot manufacturing, Tesla.
00:01:02.000 --> 00:01:07.680
Tesla says its Fremont line has begun producing its first production quality Optimus robots.
00:01:07.920 --> 00:01:13.120
It has discussed a long-term capacity target of 1 million robots a year for that line.
00:01:13.280 --> 00:01:20.079
But the latest shareholder update lists Optimus as early ramp and gives no installed capacity figure.
00:01:20.400 --> 00:01:26.480
CEO Elon Musk said it could take at least a year to reach about 20,000 robots per week.
00:01:26.640 --> 00:01:29.200
That is a target, not current output.
00:01:29.359 --> 00:01:32.560
Closer to named third-party trials, Aptronic.
00:01:32.879 --> 00:01:36.400
Apollo 2 is being used for data collection and pilot work.
00:01:36.480 --> 00:01:40.719
And Aptronic says its robots are active at customer and partner sites.
00:01:41.040 --> 00:01:49.120
Reuters reported that Aptronic had built hundreds of Apollo 2 robots by June, but the company did not disclose how many were deployed.
00:01:49.359 --> 00:01:52.400
Closer to proven production economics, neither.
00:01:52.640 --> 00:01:56.480
Aptronic's named relationships are not evidence of scaled production.
00:01:56.640 --> 00:02:09.840
For context, logistics partner GXO said it had completed 45 humanoid pilots across the industry, had not reached return on investment with humanoids, and expected zero humanoids in production at GXO in 2026.
00:02:10.159 --> 00:02:15.599
That figure is not specific to Apollo, but it's a useful reality check for the whole sector.
00:02:15.919 --> 00:02:19.520
My verdict, Tesla leads on the factory that builds the robot.
00:02:19.759 --> 00:02:22.960
Aptronic leads on visible customer and partner engagement.
00:02:23.199 --> 00:02:26.800
Neither has yet proved the robot as a dependable production worker.
00:02:26.960 --> 00:02:28.800
What versions are we comparing?
00:02:29.039 --> 00:02:40.560
As of October 29, 2026, Tesla's latest production effort is associated with Optimus Gen 3, which Tesla has described as its first design intended for mass production.
00:02:40.800 --> 00:02:44.159
But Tesla has not published a complete current spec sheet for it.
00:02:44.240 --> 00:02:51.039
The company's October update confirms early production and says the manufacturing process is suitable for a commercial product.
00:02:51.280 --> 00:02:57.360
It does not disclose the robot's detailed dimensions, payload, runtime, or current task results.
00:02:57.599 --> 00:03:05.759
Aptronic's latest disclosed platform is Apollo 2, unveiled on June 30, 2026, in both bipedal and wheeled versions.
00:03:06.000 --> 00:03:10.080
Aptronic describes it as a data collection and training platform for pilots.
00:03:10.240 --> 00:03:17.680
The company calls Apollo 3 its next generation commercial product, but has not published its specifications or affirm launch date.
00:03:17.919 --> 00:03:28.240
CEO Jeff Cardenas told Reuters that pilot work would continue through 2026, with real production versions expected in 2027 and beyond.
00:03:28.479 --> 00:03:35.680
A major caution: Aptronic's well-known numbers for height, weight, lift, and battery runtime were published for the original Apollo.
00:03:35.840 --> 00:03:39.599
They should not be treated as confirmed Apollo 2 specifications.
00:03:39.840 --> 00:03:43.360
The same caution applies to Tesla's early concept measurements.
00:03:43.520 --> 00:03:45.360
They are not Gen 3 measurements.
00:03:45.599 --> 00:03:51.439
Hardware comparison.com.
00:03:51.680 --> 00:03:57.199
The table shows a real difference in disclosure, not necessarily a difference in hardware quality.
00:03:57.439 --> 00:04:01.199
Apptronic publishes more about battery swapping and safety zones.
00:04:01.360 --> 00:04:04.960
Tesla publishes more about factory investment and production plans.
00:04:05.120 --> 00:04:10.319
Neither gives buyers the operating data needed for a full cost and reliability comparison.
00:04:10.639 --> 00:04:12.319
Autonomy and perception.
00:04:14.879 --> 00:04:20.480
Tesla's pitch is that its experience in vehicle AI will help optimists learn real-world tasks.
00:04:20.639 --> 00:04:23.040
But factory manipulation is not driving.
00:04:23.279 --> 00:04:26.800
A robot must grip, lift, place, and recover from small errors.
00:04:26.959 --> 00:04:34.399
Tesla has not published a comparable Gen 3 task success rate, cycle time, or full shift autonomy result in its latest update.
00:04:34.560 --> 00:04:37.600
The Q3 2026 call points to the remaining work.
00:04:37.839 --> 00:04:46.319
Musk said Tesla needs roughly 10,000 to 20,000 robots for an Optimus Academy, where robots can practice tasks and generate learning data.
00:04:46.560 --> 00:04:55.279
He also put outside company deliveries around the end of 2027 and said reaching about 20,000 robots per week at Fremont could take at least a year.
00:04:55.519 --> 00:04:59.600
Those are company forecasts, not confirmed dates or output levels.
00:04:59.920 --> 00:05:02.560
Apptronic has a more visible data pipeline.
00:05:02.800 --> 00:05:10.160
It says Apollo 2 gathers data through both teleoperation, a human steering or guiding the robot, and autonomous execution.
00:05:10.319 --> 00:05:15.439
Apptronic and Google DeepMind intend to use that data to improve future robot models.
00:05:15.680 --> 00:05:22.160
This is a thoughtful plan, but teleoperated data collection is not the same as autonomous production work.
00:05:22.399 --> 00:05:26.399
Apptronic does not publish the share of work done without human help.
00:05:26.639 --> 00:05:31.839
Google DeepMind has published test results for Gemini Robotics 2 on Apollo 2.
00:05:31.920 --> 00:05:43.839
In whole body tests, the robots' reported success rates included 68.4% for picking up from a table, 45.7% from the floor, and 76.3% from a shelf.
00:05:44.079 --> 00:05:49.199
In multi-finger tests using a sharp wave hand, results varied widely by task.
00:05:49.519 --> 00:05:56.720
92% for unscrewing a bulb, but 32% for a dustpan task, and 40% for sealing a bag.
00:05:56.959 --> 00:05:59.920
Google notes that movement speed still needs work.
00:06:00.160 --> 00:06:04.000
These are useful research results, not factory throughput figures.
00:06:04.240 --> 00:06:05.519
Oh, interpretation.
00:06:05.759 --> 00:06:11.120
Aptronic currently has the strenger public evidence for how an AI model handles specific robot tasks.
00:06:11.279 --> 00:06:13.920
The results also show how much remains to be done.
00:06:14.160 --> 00:06:22.240
Tesla's large vehicle data advantage may help with seeing and learning, but it does not remove the hard work of reliable grasping and physical task execution.
00:06:22.560 --> 00:06:23.680
Factory pilots.
00:06:23.839 --> 00:06:26.160
Partners are not the same as production customers.
00:06:26.399 --> 00:06:28.480
Tesla a robot production line.
00:06:28.720 --> 00:06:31.120
But limited public proof of robot work.
00:06:31.279 --> 00:06:36.480
Tesla's official materials show a long shift from pilot plans to robot manufacturing.
00:06:36.720 --> 00:06:43.279
In 2024, its investor update said Optimus was performing its first factory task, handling batteries.
00:06:43.439 --> 00:06:51.279
But by Tesla's Q4 2025 earnings call, Musk said Optimus was still in RD and that factory use was not material.
00:06:51.439 --> 00:06:58.959
Now Tesla says Optimus production has begun on Fremont's high volume lines and calls the units the first production quality bots.
00:06:59.120 --> 00:07:01.600
That is meaningful progress in making the robot.
00:07:01.759 --> 00:07:10.399
But the update gives no count of robots deployed on production tasks, no task cycle times, no reliability figures, and no productivity gains.
00:07:10.639 --> 00:07:16.079
The company's latest official table lists Fremont as early ramp and Texas as construction.
00:07:16.560 --> 00:07:18.720
Tesla's manufacturing targets are large.
00:07:18.800 --> 00:07:26.160
It has described a Fremont line designed for 1 million units a year, and a future Texas line designed for up to 10 million a year.
00:07:26.399 --> 00:07:29.199
But those are capacity goals, not current production.
00:07:29.360 --> 00:07:37.600
Tesla's own Q3 table lists no installed capacity number for Optimus, and Musk says the Fremont line still needs time to ramp.
00:07:37.839 --> 00:07:41.759
Apptronic, more named partners, but still pilot stage evidence.
00:07:41.920 --> 00:07:44.399
Apptronic has named several industrial partners.
00:07:44.560 --> 00:07:48.639
Mercedes-Benz, a commercial agreement to pilot Apollo in manufacturing.
00:07:48.800 --> 00:07:53.759
The proposed work includes bringing parts or kits to production workers and checking components.
00:07:54.000 --> 00:07:58.240
The announcement describes exploration and piloting, not a scaled fleet.
00:07:58.560 --> 00:08:03.199
GXO Logistics, an early proof of concept and warehouse testing relationship.
00:08:03.439 --> 00:08:08.480
GXO describes Apollo as one prototype among several humanoids it is evaluating.
00:08:08.800 --> 00:08:11.680
Jable, a pilot and manufacturing partnership.
00:08:11.839 --> 00:08:19.279
The announced tasks include inspection, sorting, kiting, line side delivery, fixture placement, and subassembly.
00:08:19.519 --> 00:08:22.720
Jable is also intended to help manufacture Apollo.
00:08:22.879 --> 00:08:26.959
The public announcement does not report output, uptime, or savings.
00:08:27.199 --> 00:08:31.040
Okay, Google DeepMind, an AI research partner, not a factory buyer.
00:08:31.199 --> 00:08:35.679
Apptronics says Apollo 2 data is being used to develop robotics AI models.
00:08:35.919 --> 00:08:42.080
Reuters reported that Apptronic had built hundreds of Apollo 2 robots but did not disclose deployment counts.
00:08:42.320 --> 00:08:48.480
That is evidence of hardware building experience, not proof that hundreds are working independently in factories.
00:08:48.720 --> 00:08:52.000
Apptronic says Apollo 2 is a training and data platform.
00:08:52.159 --> 00:08:54.159
The commercial product is expected later.
00:08:54.399 --> 00:08:55.279
Practical reading.
00:08:55.440 --> 00:08:58.159
Apptronic has a stronger map of named partner trials.
00:08:58.320 --> 00:09:01.440
Tesla has a stronger visible investment in making robots.
00:09:01.600 --> 00:09:06.480
Neither has released a public record of sustained factory output from a fleet of humanoids.
00:09:06.720 --> 00:09:08.320
Price and economic value.
00:09:08.559 --> 00:09:12.320
Neither company offers a public purchase price for the 2026 robot.
00:09:12.559 --> 00:09:20.159
Tesla's often-quoted$20,000 to$30,000 figure is Elon Musk's long-term estimate for Optimus at scale.
00:09:20.320 --> 00:09:24.639
It is not a Gen 3 quote, and Tesla has not disclosed the current cost to build one.
00:09:24.879 --> 00:09:27.759
Aptronic has not published an Apollo 2 price.
00:09:27.919 --> 00:09:35.679
Coverage has reported that B Capital's Howard Morgan expected Apollo orders in 2027 at roughly$80,000 per robot.
00:09:35.840 --> 00:09:40.559
That is an investor's forecast, not an Atronic price list or signed purchase order.
00:09:40.720 --> 00:09:44.000
It should not be used as a firm cost in a business case.
00:09:44.240 --> 00:09:48.559
For a plant buyer, sticker price is only one part of the decision.
00:09:48.799 --> 00:09:52.159
The useful number is cost per successful task.
00:09:52.399 --> 00:10:03.120
Include robot purchase or lease, software, integration, battery packs, replacement hands and actuators, safety changes, service, human supervision, and downtime.
00:10:03.279 --> 00:10:08.799
Then compare that total with the cost and speed of the process the robot would actually handle.
00:10:09.039 --> 00:10:16.240
A robot that is cheap but needs constant help may cost more than a robot with a higher price and better uptime.
00:10:16.399 --> 00:10:21.360
And a purpose-built machine may beat either humanoid if the task never changes.
00:10:21.600 --> 00:10:26.320
What executives and investors should watch next before approving a large deployment?
00:10:26.559 --> 00:10:30.799
Ask for results that neither company has publicly supplied in a complete form.
00:10:30.960 --> 00:10:34.080
How many hours did the robot work without a person guiding it?
00:10:34.240 --> 00:10:39.200
How many task cycles did it complete per hour compared with a human or existing machine?
00:10:39.440 --> 00:10:42.320
How often did it fail, stop, or need a reset?
00:10:42.559 --> 00:10:46.080
How long does it take to replace a battery or repair a hand?
00:10:46.320 --> 00:10:50.480
Can it keep working across a full shift under real factory condition?
00:10:51.360 --> 00:10:54.639
Ons, what is the all-in cost per completed task?
00:10:54.799 --> 00:10:58.480
For a pilot, set a narrow task and a pass-fail bar.
00:10:58.720 --> 00:11:07.039
For example, move a defined part between two known stations at a target rate, with a set limit on human interventions and downtime.
00:11:07.440 --> 00:11:11.519
Do not accept a stage demo as a substitute for a measured production test.
00:11:11.840 --> 00:11:17.360
Final verdict: Tesla Optimus is closer to factory scale manufacturing of humanoid robots.
00:11:17.600 --> 00:11:21.279
Aptronic Apollo is closer to external industrial pilot work.
00:11:21.519 --> 00:11:25.919
Tesla's Fremont Line and production quality units are the clearest scale signal.
00:11:26.159 --> 00:11:35.200
Aptronic's named partners, data collection program, swappable battery plan, and published Google DeepMind tests give buyers more to inspect today.
00:11:35.440 --> 00:11:40.159
But neither robot is ready to be treated as proven factory labor at scale.
00:11:40.399 --> 00:11:44.639
The next milestone is not a bigger factory or a more impressive video.
00:11:44.799 --> 00:11:52.720
It is a customer publishing the numbers, autonomous hours, task speed, failure rate, repair time, and cost per task.
00:11:52.960 --> 00:12:01.120
Until then, Tesla is the stronger scale bet, Haptronic is the stronger pilot bet, and both remain early stage deployment projects.
00:12:01.279 --> 00:12:05.039
All links to sources are available in the text version of this article.
00:12:05.279 --> 00:12:09.360
You can find the full article at robotcomparisons.com slash blog.
00:12:09.600 --> 00:12:10.559
Thanks for listening.
00:12:10.799 --> 00:12:12.960
Thanks for listening to Robot Comparisons.
00:12:13.120 --> 00:12:18.159
If you found this episode valuable, make sure to subscribe so you never miss a new deep dive.
00:12:18.320 --> 00:12:25.279
For the full written breakdown, detailed spec sheets, and side by side comparisons, head over to robotcomparisons.com.
00:12:25.360 --> 00:12:30.559
That's R O B O T C O M P A R I S O N S dot com.
00:12:30.879 --> 00:12:33.120
New research drops multiple times a week.
00:12:33.360 --> 00:12:34.559
See you in the next one.
