Latest / Elon Musk Podcast / Tesla’s Optimus Meets Grok: The Robot That Builds

Transcript

Tesla’s Optimus Meets Grok: The Robot That Builds · Elon Musk Podcast

• 0:01
  Hey everybody, welcome back to the Elon Musk Podcast.
• 0:05
  This is a show where we discuss the critical crossroads, the
• 0:08
  shape, SpaceX, Tesla X, The Boring Company and Neurolink.
• 0:13
  I'm your host Will Walden. Tesla quietly revealed a new
• 0:18
  generation prototype of its Optimus robot, and this one is
• 0:21
  powered by Crock, the in house AI from XAI.
• 0:25
  Now, the reveal wasn't part of some flashy thing that Elon Musk
• 0:29
  did on stage and brought out a robot.
• 0:32
  No, not at all. Came from a brief clip tucked
• 0:35
  inside a recruitment video posted on X.
• 0:39
  The footage shows a redesigned robot moving with noticeably
• 0:42
  improved balance and dexterity, assembling parts in what looks
• 0:46
  like a mock factory setting. The timing of this update raises
• 0:51
  a question. How close is Tesla to turning
• 0:54
  Optimus from a lab project into a real product that everybody
• 0:59
  could buy? Now let's talk about the video a
• 1:02
  little bit. This shows the new Optimus using
• 1:04
  both arms to pick up wires and insert them into a board with
• 1:08
  stable, repeatable motion. Never been done with Optimus.
• 1:13
  The entire motion sequence lasts just a few seconds, but it
• 1:16
  demonstrates something important.
• 1:18
  Tesla is moving away from choreographed robot stunts.
• 1:23
  No backflips or anything weird like that.
• 1:25
  And they're not trying to run up dirty hills.
• 1:28
  They're just doing things that are practical.
• 1:32
  You know, they're focusing on task automation that aligns with
• 1:35
  its manufacturing prowess in the future.
• 1:38
  So first and foremost, Optimus is going to be a production
• 1:42
  robot manufacturing 100% going into factories.
• 1:47
  That's exactly what's going to happen.
• 1:49
  Enterprise companies, big corporations will be using Tesla
• 1:55
  bots for that first. And then once that's proven and
• 1:59
  once it can achieve those tasks repeatedly, Elon and Tesla will
• 2:07
  of course, optimize Optimist for us.
• 2:09
  So it'll be able to do your dishes, it'll be able to mow
• 2:12
  your lawn. It'll be able to, I don't know,
• 2:15
  make your baby burp, Pat, it's back and make, I don't know,
• 2:19
  Optimist is pretty optimistic those things.
• 2:24
  So they confirm, Tesla confirmed that this version of Optimist
• 2:28
  runs on Grok. It's a huge shift.
• 2:32
  The original robot operated using custom software built in
• 2:35
  house with limited natural language integration.
• 2:39
  And by switching to Grok, Tesla gains access to a language model
• 2:43
  that can parse instructions, adjust to variable tasks, and
• 2:47
  potentially train itself on new workflows with fewer hard coded
• 2:51
  commands. Makes the robot very, very
• 2:54
  useful. The use of Croc also hints at a
• 2:57
  longer term plan to unify Tesla's robotics software and AI
• 3:01
  stack. This goes into SpaceX as well.
• 3:04
  These robots will land on Mars. If Croc powers Optimus Autopilot
• 3:10
  and the user interface inside Tesla's vehicles, then Tesla's
• 3:14
  future AI models won't just answer your questions.
• 3:19
  They're going to be part of physical systems.
• 3:22
  This is what makes Optimist better than other bots out
• 3:26
  there. It's a competitive advantage no
• 3:29
  other company or robotics lab has right now.
• 3:33
  Tesla's new footage didn't show Optimist doing high speed
• 3:36
  walking or climbing stairs this time, but that's not the point
• 3:40
  of this video. The video focuses on fine motor
• 3:44
  skills that matter most inside a factory and inside your homes.
• 3:49
  The robot doesn't need to run. It doesn't need to lift things,
• 3:55
  it needs to connect, hold and repeat the actions without
• 3:59
  missing a beat. It can just do the things.
• 4:02
  If they can do that consistently, it could reduce
• 4:04
  labor costs across Tesla's production lines and set a
• 4:07
  precedent for robotic integration in a high volume
• 4:11
  manufacturing. Now think about that for a
• 4:13
  second. They replace everybody on their
• 4:18
  staff that's doing these physical labor tasks, these
• 4:22
  complex, dexterous tasks, things that you have to manipulate with
• 4:29
  your fingers, adding insulation, adding a wiring, things like
• 4:35
  that that you have to do to a car to manufacture it.
• 4:39
  Those people are going to be fired.
• 4:41
  Elon is actually building the replacement for the people in
• 4:44
  the Tesla factories. Sure, there will be some people
• 4:46
  left, of course, but automation is important to auto
• 4:51
  manufacturing. And of course Elon wants to cut
• 4:55
  costs. Why would you pay people when
• 4:58
  you could just build a robot that's going to be able to
• 5:01
  manufacture something 24/7? People have to take lunch
• 5:04
  breaks. People have to go to the
• 5:05
  bathroom. People have to have meetings,
• 5:08
  you know, and there has to be safety meetings in the mornings.
• 5:12
  And everybody has to do human things.
• 5:16
  I mean, you might get a phone call in the middle of the day,
• 5:18
  your kids sick, you got to go pick him up from school.
• 5:20
  That person is gone for the day. So instead of that person
• 5:24
  leaving for the day, Elon wants to replace it with an optimist
• 5:27
  robot. This is what's happening.
• 5:30
  This is exactly what's happening.
• 5:31
  And you know as well as I do this workforce, once it's gone,
• 5:35
  the Tesla stock price will shoot through the atmosphere to the
• 5:41
  moon, to Mars and beyond. It's going to ride a Starship
• 5:45
  into the heavens. And the stock price, get it now.
• 5:49
  Get the stock now. Because once these people are
• 5:51
  replaced, Oh my Lord, they're going to save so much money.
• 5:56
  And with that, investors are going to go wild.
• 6:00
  So this is this is important. The timing of this update also
• 6:04
  lines up with Musk's recent statements about optimist
• 6:06
  reaching commercial readiness in 2025.
• 6:10
  He's talked about this before. They've already done internal
• 6:14
  testing of optimistic gigafactories.
• 6:17
  You haven't said which ones they've done or what task is
• 6:20
  actually doing, but it's working on it.
• 6:23
  Tesla employees have repeatedly been using the robot and
• 6:26
  simulated environments to evaluate how it performs under
• 6:29
  factory conditions. So think about this.
• 6:36
  I was in this situation before where you replace yourself with
• 6:40
  somebody that you train, you train them on the skill that you
• 6:45
  know, and you're the only person that knows that skill in that
• 6:47
  job. And they brought somebody in and
• 6:52
  under the the guys that you will be training them to help you.
• 6:59
  What you're actually doing is training somebody who gets paid
• 7:04
  less than you to replace you. This happened to be and it was
• 7:07
  hilarious. I already had another job lined
• 7:10
  up. I didn't like that job.
• 7:11
  So I was out anyway. So what I did was train somebody
• 7:17
  to replace me, not knowing that they were going to replace me.
• 7:20
  I thought I still had, you know, maybe like, I, I could have
• 7:22
  stayed there for as long as I wanted to, as long as I didn't
• 7:24
  replace this person. But I cost the money.
• 7:27
  I cost the company too much money, apparently.
• 7:28
  And that's exactly what they're doing with these robots.
• 7:31
  These people. The employees have been using
• 7:34
  the robot and simulated environments to evaluate how it
• 7:37
  performs under factory conditions.
• 7:40
  In a factory. If you're training a robot to do
• 7:43
  jobs that humans do, those human jobs are going to be gone in a
• 7:47
  few years. I give it five years.
• 7:50
  Every humans going to be gone from a Tesla factory, Every
• 7:53
  single person using Grok, you know, they have vision language
• 7:58
  models, they can interpret voice commands, but also images and
• 8:02
  schematics. Now it changes how a robot
• 8:04
  understands tasks. Instead of needing line by line
• 8:09
  instructions, you could just read it.
• 8:12
  Optimist reads these things. They read a wiring diagram or
• 8:15
  watch a human do a task and they remember.
• 8:17
  They learn. Then they can try to replicate
• 8:19
  it. This changes the ceiling of what
• 8:21
  general purpose robotics can accomplish, especially when
• 8:24
  paired with Tesla's vertically integrated design.
• 8:26
  Everything is made by Tesla. So optimist, you know it was.
• 8:32
  People were skeptic because Elon was overhyping it over promising
• 8:38
  timelines, and the bot was a person in a spandex suit in the
• 8:43
  first presentation. Remember that?
• 8:46
  That was pretty funny. Everyone was like, hey, that's a
• 8:49
  guy. But two years later, the company
• 8:52
  is showing a robot that can do real work with his hands.
• 8:55
  Hands are the hardest part of a robot.
• 8:59
  Think about how many bones you have in your hands, what you do
• 9:01
  with your hands. It's very complex.
• 9:04
  Same with your feet. Your feet have so many bones in
• 9:06
  them. Very complex.
• 9:08
  Now, a robot doesn't really have to move a lot usually.
• 9:12
  And if there's a a way for the robot to maneuver itself better
• 9:17
  than a human does, then that's even better for Elon and Tesla
• 9:25
  because the robot will be moving faster and in better positions
• 9:29
  than a human does. Humans get confused.
• 9:31
  You know, there's five people in front of you.
• 9:33
  What do you do when you go around them or you ask them,
• 9:35
  hey, excuse me, can I come through, please?
• 9:39
  And then the, you know, they might spread out, but a Tesla
• 9:42
  bot, they'll have patterns that they can use all through the day
• 9:45
  so nobody gets stuck. There's no stoppage, no
• 9:48
  slowdowns. And the use of Grok acts as a
• 9:54
  public test bed for XA is models now.
• 9:56
  Training a robot on Grok creates a feedback loop that can refine
• 10:00
  Grok. So the machine is learning
• 10:03
  people. Tesla has already merged XAI&X
• 10:05
  Corp into the same parent company.
• 10:07
  Bending the same AI powering tweets could now power robots to
• 10:11
  build cars. Crazy data to model to machine.
• 10:21
  This is insane system level AI integration that Musk has talked
• 10:24
  about for years now. Tesla hasn't actually said when
• 10:29
  the next optimist prototype will be shown.
• 10:31
  Again, this is just a one off and it's not even a thing that
• 10:34
  Tesla did or if it even plans a public demonstration for this.
• 10:38
  But the company is a pattern of free using the short controlled
• 10:41
  releases to test responses and also they use it to recruit
• 10:45
  people. If you see a cool project that
• 10:48
  you want to work on, of course you're going to apply to that
• 10:51
  thing. You know, I've applied to so
• 10:54
  many jobs that I wanted to be part of this crazy team that's
• 10:58
  going on, you know, this crazy project because I want to build
• 11:03
  something that's interesting to me.
• 11:04
  And if you're a robotics person or you're an AI person and you
• 11:08
  see Grok and Optimist working together a full AI suite inside
• 11:13
  of a robot body, which is also connected to the cloud.
• 11:17
  Oh my God, I think this is Terminator.
• 11:19
  I think Elon is building Terminator right now.
• 11:22
  Cyberdyne Systems I yeah, it's it's all coming to fruition.
• 11:29
  You know, we, we think it's going to be the end of humanity
• 11:31
  with these robots. But in all actuality, this
• 11:34
  utopian world that they promise us isn't actually going to be a
• 11:38
  utopian world. People will be losing their
• 11:42
  jobs. And I'm an optimist.
• 11:44
  I'm just also a realist. I want these things to happen.
• 11:49
  I want the Optimist robot to be part of everybody's life.
• 11:54
  I could use a robot. I mean, my kid talks to Alexa to
• 12:02
  brush his teeth, right? And he asks Alexa to play a
• 12:05
  song, and then he brushes his teeth along with a song.
• 12:08
  And then he'll say, thank you, Alexa, stop, please.
• 12:13
  And then Alexa will stop. And then he might ask Alexa a
• 12:18
  question, which to me is really intriguing, right?
• 12:21
  He might ask Alexa, you know, something that you know
• 12:26
  something about wherever or whatever, like a monkey that he
• 12:30
  saw on his mom's phone or something might ask a question
• 12:34
  about them. Or he might just start talking
• 12:36
  to it. And I'm like, buddy, you know,
• 12:39
  that's not a real thing, right? So he knows that it's not a real
• 12:41
  person, but it's an interesting thing, right?
• 12:45
  I used to play with toys that would talk and it's similar to
• 12:49
  that when I I had a little robot when I was a kid and it would
• 12:53
  say certain phrases. Then I talk back to it because
• 12:56
  you know, that's what you do when you're a little kid.
• 12:58
  You just find ways to express yourself.
• 13:00
  And when he's talking to Alexa, this is exactly the kind of
• 13:05
  project that Optimus is. Optimus is a robot that's using
• 13:10
  language to build robots that use the same language to build
• 13:18
  more robots and then to build things and to do things around
• 13:22
  your house. What if, and I know we already
• 13:25
  have robot lawn mowers, but what if we had a robot groundskeeper?
• 13:34
  They could trim the hedges, They could, you know, they could
• 13:39
  sweep, they could mop, they could mow your lawn.
• 13:44
  Of course, they could fix bricks that are broken and you're in
• 13:49
  your fire pit or something like that.
• 13:50
  You know what if it's something like that where they just
• 13:53
  optimize your life, Optimus optimizes your life.
• 13:57
  And of course you're not going to buy it outright.
• 13:59
  You're going to have to rent it because why would you not have a
• 14:03
  subscription for an Optimus robot unless you buy the robot
• 14:07
  and then you have subscription for the AI, for the use of the
• 14:11
  robot. You buy the hardware and then
• 14:15
  you have a subscription for the robot, which is going to be
• 14:19
  wild. But you know, yeah, upgrade
• 14:22
  packages sort of like Tesla, you know, you can upgrade your
• 14:25
  upgrade your package for a car. So it'll be similar, similar to
• 14:29
  that for Grok. Like you could have Grok and
• 14:33
  Optimus. You can have an Optimus robot
• 14:36
  that does things in the house. Then you upgrade your package to
• 14:40
  the whole house where it fixes things and handyman kind of
• 14:43
  stuff. You know, maybe if your sink
• 14:47
  breaks, you could be, hey Optimist, you know, sink is
• 14:49
  broken. Can you see what's wrong with
• 14:51
  it? Try to fix it.
• 14:54
  Send your robot in there before you hire a plumber, before you
• 14:57
  hire, before you even call a plumber.
• 14:59
  Optimist could go in there and take a look.
• 15:01
  So not only is Optimist going to, and I've tried to be bleak,
• 15:05
  I'm just trying to say things that are going to happen.
• 15:09
  What happens, though, when Optimist does eventually take
• 15:13
  over those kind of jobs, a skilled labor jobs, that's going
• 15:18
  to be wild. It's going to be interesting in
• 15:24
  the next 20 years what's going to happen with robotics, because
• 15:26
  we're at a point now with AI and robotics.
• 15:29
  If we can get to AGI and these robots can think for themselves
• 15:32
  and just do things that need to be done around the house, maybe
• 15:38
  everybody's going to be, you know, you have a task force of
• 15:42
  five people to build an addition to your house and why do you
• 15:48
  need a, you know, why do you need a Carpenter anymore?
• 15:51
  You don't, if optimist can swing a hammer, you know, it's going
• 15:57
  to be wild. It's going to be wild.
• 15:59
  I can't wait for it, but also I'm, I'm, I'm skeptical, but
• 16:04
  also optimistic about it. It's going to be interesting.
• 16:08
  Hey, thank you so much for listening today.
• 16:10
  I really do appreciate your support.
• 16:12
  If you could take a second and hit the subscribe or the follow
• 16:15
  button on whatever podcast platform that you're listening
• 16:17
  on right now, I greatly appreciate it.
• 16:20
  It helps out the show tremendously and you'll never
• 16:22
  miss an episode. And each episode is about 10
• 16:25
  minutes or less to get you caught up quickly.
• 16:28
  And please, if you want to support the show even more, go
• 16:31
  to patreon.com/stagezero and please take care of yourselves
• 16:37
  and each other and I'll see you tomorrow.