Transcriptly

This Humanoid Robot Just Got Human-Level Hands. Here's the Man Behind It | 1X, Bernt Børnich

EO17:09Added Aug 31, 2026

*This interview was filmed in May 2025.1X Founder & CEO Bernt Børnich has spent the past 10 years chasing one mission — building robots that move, learn, and...

Watch on YouTube →
Contributed by 刘嘉琪

Transcript

Transcript format

  1. Intro

  2. 00:00So I think accepting failure
  3. 00:05and having that as part of your culture
  4. 00:07is incredibly important because if if
  5. 00:09not you're not innovating right. Most of
  6. 00:10the things you do that has never been
  7. 00:12done before will just be plain out
  8. 00:14wrong. In hindsight it might not just be
  9. 00:16plain wrong. It might be like borderline
  10. 00:18stupid. You're like how did we ever
  11. 00:20think this was going to work? But that's
  12. 00:21that's just the nature of the game.
  13. 00:23That's that's how innovation works. I
  14. 00:24think the only thing that's not
  15. 00:26acceptable from a culture point of view
  16. 00:27is you didn't really try. you didn't
  17. 00:30give it everything you had or you didn't
  18. 00:32actually reflect over why did it fail so
  19. 00:34you can learn. It is how we make
  20. 00:36progress. Can you do this under
  21. 00:38pressure? Can you keep this culture
  22. 00:39where failure is okay even when there's
  23. 00:42pressure? And if you manage to do that
  24. 00:44then you get great innovation. Focus on
  25. 00:47the things that make you happy. It's
  26. 00:48going to be a long journey and if you're
  27. 00:49not having fun you're not going to make
  28. 00:51it. There's this notion of like being a
  29. 00:53founder is like grind, grind, grind, and
  30. 00:55it is and it's going to be a lot of
  31. 00:57grind and it's going to be a lot of dark
  32. 00:58days, but there has to be also a lot of
  33. 01:00fun to make it worth it. The best cheat
  34. 01:02code I have is pick a problem that
  35. 01:04really excites people. We do humanoid
  36. 01:06robots. It's really freaking cool.
  37. 01:09[Music]
  38. 01:17Thanks, Neil.
  39. 01:24So, my name is Bar Borick. I'm the
  40. 01:26founder of OnX and we make humanoid
  41. 01:27robots for the whole. We thought about
  42. 01:29it for a long time. We just got really
  43. 01:31annoyed that everyone was posting
  44. 01:32robotic videos online with like 4x, 5x,
  45. 01:356x speed because robots generally don't
  46. 01:38move naturally, don't don't move at
  47. 01:40human speeds. We just thought it was
  48. 01:41really funny like let's make sure like
  49. 01:43in every frame there's a 1x. And our
  50. 01:46mission is to create an abundance of
  51. 01:47labor through these intelligent
  52. 01:49machines. How life will be when you
  53. 01:50actually have a humanoid in your home
  54. 01:52when you can really focus on enjoying
  55. 01:54the time we spend together, right? And
  56. 01:56the things that make us human
  57. 01:59[Music]

  58. When a Humanoid Robot Becomes Your Roommate

  59. 02:03when I was a small kid. I got a hold of
  60. 02:05my first computer. Kind of like clicked,
  61. 02:07right? And that was really magical to
  62. 02:08me. You can write some code and you can
  63. 02:10have something else move in the physical
  64. 02:11world. You can combine these two. As I
  65. 02:13kind of started looking at everything
  66. 02:14around me and like how how things that
  67. 02:17move very efficiently get things done
  68. 02:20and like how much the physical world
  69. 02:21matters. I decided I want to make
  70. 02:23humanoid robots. I got really inspired
  71. 02:25by Honda Asimo back in the day. Most
  72. 02:27magical thing about Asimo was the
  73. 02:28interactions with people. It wasn't an
  74. 02:30industrial automation system. This was
  75. 02:32actually kind of straight out of Star
  76. 02:34Wars, right? It's this human robot that
  77. 02:35can walk around, run around, hand you a
  78. 02:38bottle of water and make sure that all
  79. 02:41these chores we don't want to do every
  80. 02:42day. Like we don't need to do them. We
  81. 02:44can have robots for this, right? That's
  82. 02:45really what also motivated me when I
  83. 02:47started 1x. Just look at why did Asimo
  84. 02:50actually fail. So if you look at Asimo,
  85. 02:52they followed what I would call like the
  86. 02:53classical robotics regime, the same as
  87. 02:55we see mostly in factories and work
  88. 02:57cells of how you build this. And it's
  89. 02:58not really inspired by nature and how we
  90. 03:01humans are built and how we humans move.
  91. 03:03This made it very hard for them to
  92. 03:05actually be able to do useful things
  93. 03:07outside of the lab because the system
  94. 03:09they built really had to make a lot of
  95. 03:11assumptions about the environment which
  96. 03:12doesn't work in the real world, right?
  97. 03:14Because in this creative chaos around us
  98. 03:16in everyday life that is where so much
  99. 03:18of our intelligence comes from and our
  100. 03:19lessons learned. You cannot really make
  101. 03:21that many assumptions. Things change all
  102. 03:22the time. We have to make it safe. We
  103. 03:24make have to make it affordable. But it
  104. 03:26can't be a toy. It actually has to do
  105. 03:28the work. It has to be very useful if we
  106. 03:29want these machines to be truly
  107. 03:31intelligent. Also, if we want these
  108. 03:33machines to behave in a manner where
  109. 03:34they're aligned with us, they have to
  110. 03:36live and learn among us.
  111. 03:40So, 1X was started about 10 years ago
  112. 03:42now. Really, it started out from the
  113. 03:44perspective of how can we make humanoid
  114. 03:46robots that can actually have a real
  115. 03:48impact on the world. Norway was very
  116. 03:49good in the beginning. We had very
  117. 03:51little attention. We were all by
  118. 03:53ourselves. It was just a group of people
  119. 03:55coming together, spending every waking
  120. 03:57hour figuring out how to solve the solve
  121. 03:59the problem. We moved a lot of talent
  122. 04:02over to Norway. The field at that point
  123. 04:03was very small. Thinking about like who
  124. 04:05was like really the best people in
  125. 04:07humanoid robotics back in like 2015
  126. 04:092016, right? There wasn't that many
  127. 04:10people. I remember still when like on
  128. 04:12the humanoid robotics conference we can
  129. 04:13like fit everyone in one room. This
  130. 04:15created this kind of like very exciting
  131. 04:17ecosystem where everyone was basically
  132. 04:19almost living together and just living
  133. 04:21and breathing this problem really became
  134. 04:22this great group of friends that just
  135. 04:24went on this adventure together. We were
  136. 04:26very early in this space. There weren't
  137. 04:27that many believers. It's always
  138. 04:29challenging to raise money. The space
  139. 04:31isn't really something people believe in
  140. 04:33yet. Around like 2017, I would say, is
  141. 04:35where we struggled the most. We actually
  142. 04:37had fundraising locked in. We were one
  143. 04:40signature away from maring money and
  144. 04:42then COVID hit. And that's probably the
  145. 04:44most painful thing I've done as uh as a
  146. 04:45founder. So, we went down to half the
  147. 04:47number of people to get through CO
  148. 04:49because of course in the beginning of CO
  149. 04:51there was no raising money. Everyone was
  150. 04:54just sitting on the fence figuring out
  151. 04:55how what happens now. And of course
  152. 04:59these were like people who had moved to
  153. 05:01Norway to be part of this like giving up
  154. 05:03everything they had and like they want
  155. 05:05to be part of this dream and they were
  156. 05:06all in right and having to let go of
  157. 05:08these people are of course extremely
  158. 05:09extremely painful. There are two ways we
  159. 05:11lose right either we lose velocity or we
  160. 05:13run out of money and those are the two
  161. 05:15things we need to make sure never
  162. 05:17happens.
  163. 05:19The first year was really about proving
  164. 05:21out can this be done? Really lying the
  165. 05:23foundations for a new paradigm in how we
  166. 05:25design robots. We decided to go in a new
  167. 05:27direction, right? A different paradigm
  168. 05:29in robotics. It's not that no one has
  169. 05:31ever done tendon drive systems before or
  170. 05:33like cable drives, but no one has really
  171. 05:35worked deeply enough and long enough on
  172. 05:37the problem to make it work. It's not
  173. 05:38easy to catch up because you need to
  174. 05:40sink a lot of work into this to make it
  175. 05:41work. It's also pretty exciting because
  176. 05:43it gives you a real moat.

  177. Why Building Humanoids Is a Whole New Level

  178. 05:45If there is a lot of energy when I move,
  179. 05:48then when I step on the ground, there
  180. 05:49will be a huge impact and this will
  181. 05:52disturb me. It will disturb the ground.
  182. 05:54Like, it's not a good idea. So, you want
  183. 05:55to make sure there's as little energy as
  184. 05:56in this as possible. It actually just
  185. 05:58comes back to kinetic energy. We have a
  186. 06:00very good intuition for this. We learned
  187. 06:02this in school. If a car moves twice as
  188. 06:05fast, it's not twice as dangerous, it's
  189. 06:08four times as dangerous because it falls
  190. 06:09to square. And this is also true for
  191. 06:11robotics. So if you think about the
  192. 06:13traditional industrial robots, they
  193. 06:14typically have gears that are about 100
  194. 06:17to1 gear ratio. So like if your arm is
  195. 06:19moving like this, something inside here
  196. 06:21is spinning 100 times faster. And
  197. 06:23there's just an enormous amount of
  198. 06:24energy in that rotation. So you can
  199. 06:26think about it, something here is
  200. 06:27spinning at 20,000 RPM and then when
  201. 06:29your arm hits something, this needs to
  202. 06:31immediately stop. There's no way it can
  203. 06:33immediately stop. It's going like at a
  204. 06:35blazing speed, right? It can't
  205. 06:36immediately stop. And everything we do
  206. 06:38when we interact with the world is
  207. 06:40collisions, right? Whether we're taking
  208. 06:41a step or whether I'm just touching my
  209. 06:43my watch or whether I'm picking
  210. 06:45something up, it's all collisions. And
  211. 06:47the way this is typically solved in
  212. 06:48factories is that you know exactly where
  213. 06:50things are. So you will see the robot
  214. 06:52move and it will kind of stop just
  215. 06:54before it touches the world because it
  216. 06:56needs to touch the world very slowly.
  217. 06:58And this works amazingly in factories
  218. 07:00and has kind of been a groundwork for
  219. 07:02like robotics working well over the last
  220. 07:0360 years. But if you're in a home or in
  221. 07:06a garden or whatever, you don't have a
  222. 07:08calibrated factory. So you don't know
  223. 07:10exactly when to stop and that's why we
  224. 07:12need these very low energy systems that
  225. 07:14can be safe both with respect to people
  226. 07:15but also with respect to itself and the
  227. 07:17world. You don't want your robot to
  228. 07:19damage your furniture, right? And of
  229. 07:20course you don't want the robot to hurt
  230. 07:22you. But if robots are going to be able
  231. 07:23to live and learn among us, they need to
  232. 07:24be able to explore in the world. They
  233. 07:26need to be able to learn through trial
  234. 07:27and error. And that means you just you
  235. 07:29need to be very low energy. You need to
  236. 07:30be soft. You need to be compliant. And
  237. 07:32humans are just an amazing example of
  238. 07:35this. When we did the early first
  239. 07:37deployments years ago, right? And we
  240. 07:38tried this out in homes. The thing that
  241. 07:41really struck us very early was how
  242. 07:43almost everything you do is social. So
  243. 07:46even just getting something in the
  244. 07:48fridge is a social act because likely
  245. 07:50there's someone in the kitchen. And now
  246. 07:52you need to clearly kind of communicate
  247. 07:54your intent. I'm going to go to the
  248. 07:55fridge and open it. Make sure you're not
  249. 07:56in the way. Make sure you do this in a
  250. 07:58safe manner. And you can't really
  251. 08:00separate these two problems. Like any
  252. 08:02kind of labor that you do among people
  253. 08:04is kind of social labor. Really,
  254. 08:06intelligence comes from diversity. That
  255. 08:08was a great insight that we had pretty
  256. 08:09early on. When you flip the light switch
  257. 08:12in the morning, there's light. And if
  258. 08:14not, you're pretty annoyed because
  259. 08:15humanity has mastered energy for
  260. 08:18practical purposes in everyday life.
  261. 08:19It's just abundant. And this same thing
  262. 08:21is going to now happen to physical
  263. 08:23labor. And this is really needed because
  264. 08:25there's not enough people getting born.
  265. 08:27We don't have enough people to take care
  266. 08:28of our elderly. And we see like prices
  267. 08:30of goods and services increasing.
  268. 08:32Everything is inherently limited by our
  269. 08:34ability to effectively serve labor. The
  270. 08:36question then becomes how do we get
  271. 08:38there as quickly as possible. So most of
  272. 08:40the human robotics companies you see
  273. 08:42today they are more component
  274. 08:44integrators buying off the shelf and
  275. 08:46integrating into a system. And this in
  276. 08:48itself is pretty hard. But the journey
  277. 08:50we set out on here with the tenant
  278. 08:52drives and our unique motors and
  279. 08:54everything else. This means that we have
  280. 08:55to do everything oursel because these
  281. 08:57components don't exist. So we spent the
  282. 08:59last 10 years right building not only
  283. 09:02the foundational technology but also
  284. 09:04actually the machines that can build
  285. 09:06these components and automation
  286. 09:09equipment and everything needed to build
  287. 09:11a factory. It's going to be extremely
  288. 09:14challenging because manufacturing always
  289. 09:16has but we've done a very good job in
  290. 09:18simplifying the product. Make sure you
  291. 09:21don't have anything that requires
  292. 09:22special alloys. Make sure your product
  293. 09:24is very light so you don't need that
  294. 09:25much material. Just simplify, simplify,
  295. 09:27simplify. Minimize part count through
  296. 09:29this. Reduce this from something that
  297. 09:31has the complexity of a car to something
  298. 09:33that gets closer to having the
  299. 09:34complexity of let's say like some kind
  300. 09:37of electrical appliance in your house,
  301. 09:38right? And the way we do that is just
  302. 09:40building it all oursel having our
  303. 09:42engineers sit in the factory make sure
  304. 09:45design, manufacturing, automation,
  305. 09:47everyone's in one room, really create a
  306. 09:49very, very efficient process.
  307. 09:53This is back in 2018. I was at a stage

  308. In This Industry, Failure Comes First

  309. 09:56and this was the old robot. He was one
  310. 09:59of the first prototypes we had and we
  311. 10:01were opening this health conference
  312. 10:02where we're talking about robots in
  313. 10:04healthcare in the long term. And I'm
  314. 10:06standing on the stage together with my
  315. 10:07robot and exactly well timed as I say
  316. 10:10safe, the robot decides to accelerate
  317. 10:13backwards, hit the wall and then face
  318. 10:16plant next to me. Bring with it like all
  319. 10:18the balloons, those in the back and just
  320. 10:22it just looks like it had a really rough
  321. 10:23night like sleeping in the balloons. I
  322. 10:26spent a lot of these things up through
  323. 10:27the years. things don't always go the
  324. 10:29way you planned. So, I think accepting
  325. 10:32failure and having that as part of your
  326. 10:34culture is incredibly important because
  327. 10:35if if not, you're not innovating, right?
  328. 10:37Most of the things you do that has never
  329. 10:39been done before will just be plain out
  330. 10:41wrong. In hindsight, it might not just
  331. 10:44be plain wrong. It might be like
  332. 10:45borderline stupid. You're like, how did
  333. 10:47we ever think this was going to work? As
  334. 10:49you get more knowledge, but that's
  335. 10:51that's just the nature of the game.
  336. 10:52That's that's how innovation works. So,
  337. 10:54I think first of all, like foster a
  338. 10:55culture where failure is okay. I think
  339. 10:57the only thing that's not acceptable
  340. 10:58from a culture point of view is you
  341. 11:00didn't really try. You didn't give it
  342. 11:02everything you had or you didn't
  343. 11:04actually reflect over why did it fail so
  344. 11:06you can learn. If you give it everything
  345. 11:08you have and you learn from your
  346. 11:09failures, then you should embrace and
  347. 11:11celebrate failure, right? It is how we
  348. 11:13make progress. And this is hard to do
  349. 11:15actually. It's it sounds like it's the
  350. 11:17cliche, right? And it sounds like
  351. 11:19something that should be pretty
  352. 11:20straightforward. But of course, we're
  353. 11:22under a lot of pressure to make this
  354. 11:23happen and we need to deliver. We have
  355. 11:25timelines. We have schedules. We need to
  356. 11:27hit our manufacturing milestones. That's
  357. 11:29really where the importance of this
  358. 11:32comes in. Can you do this under
  359. 11:34pressure? Can you keep this culture
  360. 11:36where failure is okay even when there's
  361. 11:38pressure? And if you manage to do that,
  362. 11:40then you get great innovation.

  363. How Robots Learn

  364. 11:43Let me go through like how does the
  365. 11:44robot learn in general? How does it
  366. 11:46actually work? It begs the question, how
  367. 11:47do you get to a system that's
  368. 11:48intelligent enough that when you ask it
  369. 11:50to go and get a Coke in the fridge, it
  370. 11:52at least manages to do it sometimes? And
  371. 11:54we need to bootstrap. So you start with
  372. 11:56internet data of course because we have
  373. 11:57a lot of it. Then you have some
  374. 11:59synthetic simulated data and then you
  375. 12:01need some robot data and to get that
  376. 12:03robot data we typically use
  377. 12:05teleoperation. So that means we have a
  378. 12:07human that actually embodies the robot
  379. 12:09and you see through the eyes of a robot
  380. 12:10and a robot moves like you move and it's
  381. 12:12a pretty magical experience. It's kind
  382. 12:13of like hey my hands are somewhere else
  383. 12:15and I can do something anywhere in the
  384. 12:17world and it's a very nice way of
  385. 12:18transferring knowledge from a human into
  386. 12:20a machine. And once you have a bit of
  387. 12:22this, you're now able to do these tasks
  388. 12:23autonomously. And from there, you can
  389. 12:25kind of like iterate on it and learn
  390. 12:27from the real world. In the end, it's
  391. 12:29about having enough of these robots out
  392. 12:31in the real world. Learning from just
  393. 12:33trying things. You have to be able to
  394. 12:34experiment. You have some hypothesis
  395. 12:36about how to do something. You try. You
  396. 12:38see how it went, you try again. That's
  397. 12:39how we learn, right?
  398. 12:41[Music]

  399. Build Before the Market Exists

  400. 12:44Traditionally, scale does not happen
  401. 12:47first in enterprise. And this might be
  402. 12:49slightly surprising, but if you look at
  403. 12:51the history of highly innovative
  404. 12:53products, they almost never happen in
  405. 12:56enterprise first. They happen in
  406. 12:57consumer. And this is just because
  407. 12:59enterprise is risk averse and there are
  408. 13:02just too much red tape and barriers,
  409. 13:04right? From IT departments to labor
  410. 13:06unions to risk averse CEOs, it just
  411. 13:09takes a lot of time. While if you have
  412. 13:12good product market fit, nothing scales
  413. 13:14the way consumer adoption does. This is
  414. 13:17true for a lot of products. There's a
  415. 13:18very good example of this is chat GPT
  416. 13:21and the roll out of digital AI. Open AAI
  417. 13:23really tried in enterprise for a long
  418. 13:25time and they couldn't really get it
  419. 13:26working and then they released chat to
  420. 13:29see what people do with it. People
  421. 13:30figure out this incredible diverse set
  422. 13:32of like tasks they can do with this tool
  423. 13:34and they bring it to work and now you
  424. 13:36would think that then finally you
  425. 13:37succeeded but you don't. Now enterprise
  426. 13:39actually says no no you can't use this
  427. 13:40here. This is new and dangerous. We
  428. 13:42can't do this. After a while, people
  429. 13:43actually get so annoyed that they kind
  430. 13:45of say, "Oh, if I can't use it here,
  431. 13:47I'll go work somewhere else where I can
  432. 13:48use it because this tool makes me so
  433. 13:50productive." And now you created so much
  434. 13:51bottom-up pressure that you're forcing
  435. 13:53adoption. And now you can start top down
  436. 13:55and do great in enterprise. But you have
  437. 13:57to create that forced adoption through
  438. 13:59bottom-up pressure. Juno robotics is no
  439. 14:01different. If you want to do this in the
  440. 14:02next few years, not in the next few
  441. 14:03decades, you have to go through
  442. 14:05consumers. You have to find your early
  443. 14:07adopters and your true believers that
  444. 14:09can really help you push this forward
  445. 14:10and be part of this journey. Then of
  446. 14:12course it will be used in enterprise
  447. 14:14because it's going to greatly improve
  448. 14:15your productivity. But it has to happen
  449. 14:17in consumer first.
  450. 14:20If you run a deep tech company, you're

  451. Lessons for Deep Tech Founders

  452. 14:22not really sitting that close to your
  453. 14:23customers in the beginning. You're
  454. 14:25trying to solve this fundamental problem
  455. 14:27that if you solve it, your market is
  456. 14:29there. It's like if you cure cancer,
  457. 14:31you're not wondering whether or not
  458. 14:33people will actually buy your product.
  459. 14:35And if you figure out a way to take
  460. 14:37energy and turn it into any kind of
  461. 14:39labor, any kind of product and service,
  462. 14:41clearly your market will be there. It's
  463. 14:43more a question of can you actually
  464. 14:44solve the problem. And of course, you
  465. 14:45get into product market fit questions
  466. 14:47once you start to deploy this and how
  467. 14:49you make it like gradually useful
  468. 14:51because the problem you're working on is
  469. 14:53such a big problem that you can't really
  470. 14:55just say like I'm going to just sit here
  471. 14:56in my lab and work on this for 10 years
  472. 14:58and then I'm going to go out and like
  473. 15:00launch this product. First of all, you
  474. 15:02of course want it to be aligned with
  475. 15:05humans and how we want our robots to
  476. 15:07behave around around us. But also, of
  477. 15:10course, you need to show it's useful.
  478. 15:11You need to create revenue. You need to
  479. 15:13build a business because this is a
  480. 15:14really long journey. But fundamentally,
  481. 15:17it is a deep tech type journey where
  482. 15:20it's more about solving the problem than
  483. 15:21really ensuring you sit close to your
  484. 15:23customers. And once you solve the
  485. 15:25problem, then you can start caring about
  486. 15:28how all those small details, right? that
  487. 15:31takes this from being just a solution
  488. 15:33technically to something that's like
  489. 15:34packaged as a great product.
  490. 15:39We are just starting already this year
  491. 15:42hopefully we're going to have something
  492. 15:43that is very useful but in the next few
  493. 15:45years this will change the way we live.
  494. 15:48Focus on the things that make you happy
  495. 15:51because it's going to be a long journey
  496. 15:53and if you're not having fun you're not
  497. 15:56going to make it. I think that that's
  498. 15:57that's undervalued, right? Because
  499. 15:58there's there's this notion of like

  500. Build Something That Excites People

  501. 16:00being a founder is like grind grind
  502. 16:02grind and it is and it's going to be a
  503. 16:04lot of grind and it's going to be a lot
  504. 16:05of dark days where you just need to like
  505. 16:07pull it together and just get it done.
  506. 16:09But there has to be also a lot of fun to
  507. 16:11make it worth it. And uh that just means
  508. 16:14work on the interesting problems, right?
  509. 16:15The best cheat code I have is pick a
  510. 16:17problem that really excites people. If
  511. 16:19you want to find the best people in the
  512. 16:21world to come work at something, do
  513. 16:22something that excites people. We do
  514. 16:24human robots. It's really freaking cool.
  515. 16:26And most people come in and they like
  516. 16:28say hi to a robot and they're like, "Oh
  517. 16:29man, kind like I want to be part of
  518. 16:31this. I want to do this." So whatever it
  519. 16:33should be, right? Do something that
  520. 16:34matters. As long as I get to do this, I
  521. 16:37think I'll be pretty happy.
  522. 16:46[Music]
  523. 16:56[Music]
  524. 17:01la.
  525. 17:07Wow.
This Humanoid Robot Just Got Human-Level Hands. Here's the Man Behind It | 1X, Bernt Børnich — Transcriptly