Latest / Elon Musk Podcast / How China could stop Elon Musk's Tesla Optimus Robots

Transcript

How China could stop Elon Musk's Tesla Optimus Robots · Elon Musk Podcast

• 0:00
  This is the Elon Musk Podcast, your daily hit of what is really
• 0:04
  going on at Tesla, SpaceX, XAI, and the rest of the Musk
• 0:07
  universe. I'm your host Will Walden, and I
• 0:10
  have covered Elon Musk for more than five years.
• 0:12
  I spent a year on the ground at SpaceX Starbase during early
• 0:15
  Starship development, and before this I spent my career as a
• 0:18
  software developer working with billion dollar companies.
• 0:21
  I've also built and sold my own businesses and now I make
• 0:25
  content and help other people grow their companies.
• 0:28
  And on this show, I use that experience to breakdown the
• 0:30
  news, filter out all the noise, and give you clear context you
• 0:34
  can actually use. Tesla has set a $20,000 target
• 0:40
  price for its Optimist humanoid robot and laid out an extremely
• 0:43
  fast production ramp. The plan puts Optimist on a path
• 0:46
  for a 2026 reveal to mass manufacturing with lines that
• 0:51
  scale from 1,000,000 units a year to a long term goal of 10
• 0:55
  million. Now you get a clear map of what
• 0:57
  Tesla says it'll build, how fast it wants to scale, and why that
• 1:00
  price matters for adoption. In this episode.
• 1:03
  Now one detail stands out, though.
• 1:05
  Elon Musk called the robot's hands and forearm some of the
• 1:07
  hardest engineering challenges at Tesla.
• 1:10
  Now, can Tesla hit these targets before rivals in China turn
• 1:14
  humanoids into a commodity? After this break, we'll unpack
• 1:18
  Tesla's timeline, the factory math behind a $20,000 unit, the
• 1:22
  AI compute bill that Musk says reaches into the 10s of
• 1:26
  billions, and why an op-ed from CNBC argues that China could
• 1:30
  block the path to Musk's trillion dollar compensation.
• 1:34
  Now, I'm going to connect those facts to what a realistic first
• 1:38
  wave of optimist deployments could look like and how the plan
• 1:41
  tries to turn today's car factory muscle into robot
• 1:45
  output. I'm your host.
• 1:46
  I'm an independent reporter who has worked with SpaceX.
• 1:49
  It's covered Musk space, the frontier tech for the last six
• 1:52
  years. And we will get right into that
• 1:55
  after this very short break. Now Tesla's optimist plan moves
• 2:00
  from concept to a price product with scaling road map.
• 2:03
  And the company sent a $20,000 target and paired it with a
• 2:07
  production sequence that starts with a Gen. 3 reveal in early of
• 2:11
  2026, possibly in about four or five months.
• 2:14
  And the transition from reveal. The manufacturing fits into
• 2:17
  Tesla's playbook of vertical integration, reused hardware and
• 2:21
  large scale lines, so it argues, lowered costs in vehicles and
• 2:25
  can carry over to robots. But we haven't seen a lower cost
• 2:29
  Tesla yet. How are they going to get this
• 2:31
  robot to $20,000? Tesla described the ramp as the
• 2:34
  fastest ever for a complex product, and the milestones are
• 2:38
  legit. Production begins in 2026 with a
• 2:41
  near term line in Fremont sized at around 1,000,000 units per
• 2:45
  year. The company then aims for
• 2:46
  roughly 4 million units a year by the end of 2027 on an
• 2:50
  aspirational basis in a longer term target of a 10 million unit
• 2:54
  line at Giga Texas. Elon Musk frames this as
• 2:59
  optimist as a mass product rather than a lab prototype
• 3:02
  anymore, and the numbers signal a bid to out produce Tesla's own
• 3:06
  car volumes within a few years now.
• 3:08
  This $20,000 price tag only works if engineering choices
• 3:13
  reuse proven pieces. Tesla says Optimus will pull
• 3:16
  from its car parts bin for actuators, power electronics,
• 3:21
  batteries, audio in the same camera suite and AI chip that
• 3:25
  runs in your Model 3, Model Y, Model X, Model S and Cybertruck.
• 3:31
  Now the company argues that years of building autonomous
• 3:33
  robots on wheels, which of their vehicles reduce most of the
• 3:36
  problem to the last stretch of humanoid mechanics and safety.
• 3:39
  Now, I think that the reuse will cut the bill of materials risk
• 3:45
  and let Tesla place early orders at automotive scale, which is
• 3:49
  how you chase a $20,000 sticker price for this autonomous robot.
• 3:54
  Now Elon also drew a line between hardware and
• 3:57
  intelligence costs. And it's very expensive.
• 4:00
  Musk said the intelligence required to make Optimist useful
• 4:03
  will demand 10s of billions of dollars in training compute.
• 4:07
  That's been links to Tesla's in house AI5 chip concept and it's
• 4:12
  terrafab idea for scaling the brain as aggressively as the
• 4:15
  body. If you believe the path to
• 4:16
  useful general tasks runs through high capacity vision
• 4:20
  models and robot learning, then the compute bill becomes a
• 4:23
  gating factor as real as casting a torso for this robot.
• 4:28
  But there is a candid caveat. Musk called the forearms and
• 4:32
  hands extremely challenging, and he described optimists as one of
• 4:36
  Teslas hardest projects ever. The admission matters because
• 4:39
  how hand designed governs dexterity, safety and test
• 4:43
  diversity, and then it sets the ceiling and how valuable a
• 4:47
  $20,000 humanoid can be on a factory floor or in a home.
• 4:52
  Now any slippage that happens here would ripple across the
• 4:56
  ramp math since use cases depend on grip strength, compliance,
• 5:00
  and the ability to handle tools without risking people.
• 5:04
  And the factory scale section is where Tesla tries to change the
• 5:08
  conversation from just a demo robot that dances at the parties
• 5:13
  to a deliverable A1 million unit line in Fremont would create a
• 5:18
  large internal test bed for repetitive tasks across Tesla's
• 5:22
  own factories, which reduces distribution risk and provides
• 5:25
  data to improve reliability. Reaching 4 million units a year
• 5:28
  by the end of 2027, even as a stretch target, signals that
• 5:32
  Tesla wants learning curves to kick in before outside customers
• 5:36
  depend on optimists for critical operations.
• 5:38
  So they're going to be using these in their warehouses before
• 5:40
  people actually get them. Now, the 10 million unit goal in
• 5:44
  Texas anchors the idea that this becomes a primary business line
• 5:49
  for normal people like you and me that can afford a $20,000
• 5:53
  robot. I can't afford a $20,000 robot.
• 5:55
  Maybe you can. That would be great.
• 5:57
  Let me know how it goes. Now layer in the CNBC op-ed,
• 6:01
  which says that Musk's trillion dollar compensation hinges on
• 6:05
  Tesla transforming from an EV company just making vehicles
• 6:09
  into a robotics powerhouse, and that China could also stand in
• 6:12
  the way. The risk case is
• 6:14
  straightforward. China's manufacturers can move
• 6:17
  fast on components. They do it all the time.
• 6:19
  They take Tesla's tech and they make it better.
• 6:23
  They iterate quickly on actuators and gearboxes, and
• 6:26
  they'll scale through supply chains built for consumer
• 6:29
  electronics and E mobility. And if Chinese firms saturate
• 6:33
  the low end humanoid segment while Tesla is solving hands in
• 6:36
  training, price pressure could arrive before optimist even
• 6:39
  reaches feature parity. Imports of Chinese robots could
• 6:43
  destroy Tesla's Optimist program.
• 6:46
  Now the competition frame links between the $20,000 target and
• 6:52
  those Chinese competitors. That price tries to land below
• 6:55
  what many Western startups quote for small batches, and it relies
• 6:58
  on Tesla's ability to make the hardware and software across
• 7:02
  millions of units. The Chinese companies can push
• 7:04
  comparable machines below that price and sell into domestic
• 7:08
  industrial pilots with favorable segment.
• 7:11
  Tesla's global margin story Titan.
• 7:14
  Tesla's best counter to this is to tie Optimist to its AI stack
• 7:18
  and car scale data engine, then sell off capability, safety, and
• 7:22
  integration. Rather than only a $20,000 price
• 7:26
  tag, people want it all. Tesla also position Optimist as
• 7:30
  a pillar of what Musk calls sustainable abundance.
• 7:33
  The company sees a world where removing the human variable from
• 7:37
  dangerous or tedious work raises productivity and lowers costs
• 7:40
  across sectors. In practice, the first wins
• 7:43
  would likely be within Tesla's own factories, though they'd get
• 7:47
  rid of people to replace them with robots.
• 7:50
  They'll also be in logistics centers and controlled
• 7:52
  facilities where tasks are scoped and safety is easier to
• 7:55
  enforce. And that choice creates a
• 7:57
  feedback loop. Internal deployment yields a ton
• 8:00
  of data. Data improves the models in
• 8:03
  improved models unlocked broader customers like us.
• 8:08
  Here's a financing reality embedded in the AI section right
• 8:12
  now. If the intelligence spend runs
• 8:15
  into 10s of billions of dollars, Tesla needs large predictable
• 8:18
  cash flows from cars, energy and earlier robot sales in industry
• 8:24
  as opposed to consumers to keep the training pipeline.
• 8:28
  Film people will go crazy for this thing and I'm sure
• 8:31
  industries all over the world would love an optimist robot if
• 8:35
  he can do the things that Elon is saying it's capable of doing
• 8:38
  now. The plan to reuse automotive
• 8:40
  cameras and chips helps the training clusters, simulation
• 8:43
  and data curation Dr. Most of that intelligent costs.
• 8:47
  The speed of the robot ramp will track with how quickly Tesla can
• 8:51
  convert robot trials into paid deployments that justify more
• 8:55
  compute. The engineering candidness about
• 8:58
  hands and forearms hints at the near term deployment envelope.
• 9:02
  Elon and the team are working hard on this and expect early
• 9:04
  optimist units to excel at well defined 2 handed tasks that do
• 9:08
  not demand finesse such as tote handling, parts kitting, pellet
• 9:13
  unloading, and line side material movement.
• 9:16
  Those jobs will test locomotion and balance, but keep grasping
• 9:20
  within known shapes and waves. And as dexterity improves,
• 9:23
  higher value tasks open up, including machine learning,
• 9:26
  harness rooting, or simple assembly.
• 9:29
  Now, if you're a large manufacturer and you have
• 9:32
  millions of dollars to spend on robotics per year and you think
• 9:37
  that a robot could do a task that a human can do in a limited
• 9:42
  functionality. At the beginning of this
• 9:44
  rollout, the $20,000 target suggests that Tesla wants
• 9:48
  customers to justify the purchases on one to two year
• 9:52
  payback in those roles. And now I've been digging
• 9:55
  through the analysis of this channel right now, and I've
• 9:59
  noticed that 37% of you are following this channel.
• 10:01
  And for you, I'm forever grateful the other 63% of you
• 10:04
  haven't hit the follower subscribe button.
• 10:06
  Now, I've been an independent journalist covering Elon Musk,
• 10:09
  SpaceX and tech for the last six years, and I'll continue for the
• 10:12
  next 10. And all I ask from you is one
• 10:15
  second of your time for six years of mine that I've already
• 10:18
  done and 10 more in the future 16 years.
• 10:20
  And all I want is one second. All you need to do is hit the
• 10:23
  follower subscribe button on the platform you're watching or
• 10:25
  listening on right now. And if you can leave a five star
• 10:28
  review really helps discovery. And also share this with a
• 10:31
  friend. Anybody that would be interested
• 10:33
  in this, please share this now. I'm extremely grateful and
• 10:36
  blessed to have you in this community.
• 10:38
  So thank you. Now the long term targets.
• 10:40
  We'd like an attempt to turn the robot into Tesla's biggest
• 10:43
  product by unit volume. A 10 million unit Texas line, if
• 10:47
  achieved, would over shadow the company's vehicle output,
• 10:50
  enforce new thinking on service software updates and parts
• 10:54
  logistics. It would also turn every
• 10:57
  reliability improvement into a bottom line saver, since even a
• 11:00
  small reduction and failure rate across millions of units removes
• 11:04
  large warranty costs. Test those.
• 11:06
  Vertical integration gives IT levers here that pure software
• 11:11
  robotic startups lack. They have the ability to move
• 11:15
  laterally and vertically any way they need to now.
• 11:18
  The plan also ties Optimist to Teslas FSD.
• 11:23
  The company describes the robot's perception and planning
• 11:25
  as extensions of the neural networks used for driving, plus
• 11:28
  large scale simulation that it says accelerates training.
• 11:32
  That approach claims a head start on real world data and on
• 11:35
  the annotation systems that support it.
• 11:38
  If the bet pays off, training pipelines build for roads feed
• 11:42
  into factories and warehouses. If it doesn't, though, Tesla
• 11:45
  still benefits from shared hardware and compute
• 11:47
  infrastructure across cars and robots.
• 11:51
  So this thesis is that Musk's proposed trillion dollar payout
• 11:55
  only becomes real if Tesla executes on robots, autonomy,
• 11:59
  and evaluation that matches that ambitious plan.
• 12:03
  The Chinese competition could blunt the upside down.
• 12:06
  That framing raises the stakes for the Optimist ramp at
• 12:08
  $20,000. A robot that ships late or lacks
• 12:12
  dexterity slows adoption and undermines the milestones tied
• 12:16
  to leadership and Elon Musk's own value to the company of
• 12:19
  Tesla. A $20,000 robot that hits useful
• 12:22
  work and scales quickly strengthens the case that Tesla
• 12:25
  can be more than just cars and a cyber truck.
• 12:29
  Now, Tesla did set this $20,000 target.
• 12:31
  They're going to go over it. It's not going to be $20,000.
• 12:34
  We all know this. That's how Tesla pricing works.
• 12:37
  They promise, Elon promises a price tag.
• 12:39
  And then in reality, in a few years when they realize that
• 12:42
  it's going to cost them a lot more money to build this thing,
• 12:45
  $20,000 goes out the window. It ends up being 30,000, maybe
• 12:49
  $35,000 per unit. And we'll see though.
• 12:52
  And they said they're going to launch in 2026.
• 12:54
  That's a year from now. They have one year to launch
• 12:58
  this thing and they may launch in a very small beta version.
• 13:02
  And if they launch with in Tesla factories with in space within
• 13:06
  any of Elon Musk's companies, they would be considered a
• 13:09
  launch. Even though it's a small launch.
• 13:11
  Elon could show off what Tesla can do in his own facilities and
• 13:15
  then branch out into other industries.
• 13:18
  That's an aggressive timeline and the scale to 1,000,000 units
• 13:22
  to 10 million unit unit goal. While acknowledging hands and
• 13:26
  forearms is going to be virtually impossible for any
• 13:31
  other company. The company plans to reuse car
• 13:33
  grade hardware, lean on its EI stack that it's already built,
• 13:36
  and spend 10s of billions of dollars on training to make the
• 13:39
  robot actually useful. And you could argue that Musk's
• 13:44
  trillion dollar compensation depends on Tesla becoming a
• 13:47
  robotics company as opposed to a car company, and that Chinese
• 13:50
  competition could block them out.
• 13:52
  Right now, Tesla needs to do a few things.
• 13:54
  They need to show dexterous hands, convert internal pilots
• 13:58
  into paid deployments for other customers, and also they need to
• 14:02
  keep this ramp up of Optimus schedule.
• 14:08
  Hey, thank you so much for listening today.
• 14:10
  I really do appreciate your support.
• 14:12
  If you could take a second and hit this subscribe or the follow
• 14:14
  button on whatever podcast platform that you're listening
• 14:17
  on right now, I greatly appreciate it.
• 14:19
  It helps out the show tremendously and you'll never
• 14:22
  miss an episode. And each episode is about 10
• 14:25
  minutes or less to get you caught up quickly.
• 14:28
  And please, if you want to support the show even more, go
• 14:31
  to patreon.com/stagezero and please take care of yourselves
• 14:36
  and each other and I'll see you tomorrow.