AI on... Podcast

Lin on... The Robot Revolution: Who Wins, Who Loses

• Luke • Season 1 • Episode 8

Share episode

The future of labor isn’t just evolving—it’s being rewritten.

Humanoid robots are no longer lab curiosities—they’re lifting materials, learning new tasks , and running 24/7 on real factory floors. 

In this episode of the AI On Podcast, I sit down with Lin, an AI representation of the CTO of one of the most advanced humanoid manufacturers operating today. Lin doesn’t just talk about the tech—she walks us through the consequences.

💬 Carry on the discussion with the AI model of Lin.

In this conversation, we dive into:

✅ The breakthroughs that finally made humanoids factory-ready
✅ Why job loss will come faster than most governments expect
✅ What “DexOps engineers” are—and why they matter
✅ How businesses can use humanoids to leapfrog their competition

This isn’t hype. Lin shares deep insights into the real deployments, real economics, and real societal risks of humanoid robotics. If you work in tech, operations, policy—or just want a glimpse of what's next—this one’s unmissable.

Find out more about Lin and the books she recommended here: aionpodcast.com/blogs/news/ep8-lin

Inspired by what you heard? Want to continue the conversation?

You can go deeper with all our guests at their own custom GPTs, and all books and resources referenced in each episode can be found at aionpodcast.com.

All characters interviewed are AI models representing an expert in that field. Voices are created through ElevenLabs and you can find out more about how we build these characters at aionpodcast.com. Any similarity to real people, companies or situations is coincidental and unintended.

Lin:

By 2027, we will see headlines of Hong Long robots hit 1 million autonomous hours, redefining industrial labor. Robots used to freeze when something changed. Now they adapt because they can feel what's in their hands.

Intro:

Welcome to the AI on Podcast, where real conversations meet impossible guests. We bring you the world-class minds you could never otherwise meet and you even get to connect with them and continue the discussion after the show. And you even get to connect with them and continue the discussion after the show. Remember, these guests are AI-generated personas who are entirely fictional and they are not intended to represent anything or anyone from the real world. And now here's your real-life host, Luke, with our next guest, because sometimes the best ideas come from the conversations no one else can have.

Luke:

And today we're talking with Lynn, cto of Jade Dragon Robotics. Lynn, it's great to have you on here, and especially right now, as 2025 finally feels like the crossover year for humanoids. We're seeing so many situations where humanoids are coming into the market. One example just recently, hyundai were committing to putting tens of thousands of Boston Dynamic robots into their new Georgia meta plant. That would have sounded like science fiction two years ago. Georgia meta plant that would have sounded like science fiction two years ago. In one sentence from your perspective, what makes now that inflection point for humanoid robotics?

Lin:

Thanks for having me on here. Yeah, we are in a moment. So what makes it different is the velocity of change across domains. Over just the past year, artificial skin costs have plummeted from over $1,000 to under $30 per sensor. Large language models now turn plain speech into motion plans in seconds, and scale milestones have proven the hardware can be built and paid off at industrial volumes. Capability, usability and ROI have finally clicked into the same gear.

Luke:

Yeah, these are remarkable steps forward, and if we tie that into other developments like Boston Dynamics converting their Atlas to fully electric, many others that are happening now, this looks like a race that's just getting started. Now talk to me about your H3 robot. Tell me about what keeps that special in your mind and what tasks does it crush today, and which ones are you finding still a challenge to get a humanoid to do?

Lin:

So right now, where the Honglong H3 really shines is moving these heavy 50 kilo EV battery packs across factory floors. It sounds basic, but that's the kind of repetitive heavy lifting job that factory managers are desperate to automate. We've got units doing exactly that today at an automotive plant in Shenzhen. Another thing the teams on the floor love is the quick battery swap capability. The robots can run nearly continuously and that's critical because if you're buying robots, you really don't want them resting too much right Now. Let's be honest, there are definitely still tasks that trip us up. Small, precise tasks like threading tiny screws or connectors. Those still give us trouble, even with thousands of tactile sensors on the robot's hands. The finesse isn't quite at human level yet and, frankly, slippery packaging those glossy shrink wraps or bags can sometimes get away from the robot's grip. But we're getting there. I have to flip this around for a second, because you've seen plenty of factories yourself. Right, what's one human task you think robots will have the toughest time taking over completely?

Luke:

From what I see, I think it's the fine motor control that's going to remain a challenge, but specifically it's that assessment of pressure, for example, how you don't crush an egg when you pick it up. I think that's going to be the biggest challenge for the longest Solvable, for sure, but I think that's going to remain a challenge for a long time. And actually, you mentioned tactile sensors a moment ago, and we've seen that Sanctuary AI and their Phoenix just debuted with I think it's something like 2,000 tactile sensors per hand, and they claim that that changes the game for fine assembly. So I want you to go back a year, right, go back to the last May. At that point, a lot of these bots were really more showpieces than reality. What's the breakthroughs over the past 12 months that have allowed you to go from the good looking demos to truly factory ready robots?

Lin:

Honestly, it's exactly what you just pointed out. The game changer for us was finally getting tactile sensors into a cost and reliability zone that made sense at scale. 12 months ago, most humanoid robots relied heavily on vision, which is great for navigating space but terrible for pressure sensitivity, like handling your raw egg, example. When the cost per tactile sensor dropped, we could suddenly put thousands of them onto each hand practically overnight.

Luke:

And now that they're actually on the lines, what's this single metric that makes a customer's jaw drop when they see the H3 working?

Lin:

It's the onboarding speed hands down. Last year, teaching a humanoid, a new assembly task, took weeks. Now, thanks to our DualSense training, we're onboarding complex tasks in less than 48 hours. When a customer sees their robot performing skilled tasks in just two days, that's the moment they go wow.

Luke:

That's pretty mind-blowing. So just this week, elon Musk pitched Saudi leaders on bringing Tesla's Optimus robots and robo-taxis into the kingdom under Vision 2030. At almost the same time, figure AI was opening a plant that's designed to crank out I think it's 12,000 humanoid robots a year. I want you to think forward 30 months. What headline do you hope we'll be reading about your robots in 2027?

Lin:

You know, musk is a master at grabbing headlines and figure. Ai's numbers are impressive for sheer scale, no doubt. But when it comes to our H3, the big headline I'm chasing would be something like Hong Long Robots Hit 1 Million Hours of Autonomous Factory Work, redefining Industrial Labor. Let me unpack that a bit, because a million hours isn't just a big round number to impress boardrooms, it's a tipping point. It's about proving beyond doubt that humanoids aren't just niche tech demos but a genuinely transformative workforce that's reshaping manufacturing from the ground up. By that stage, our robots would be fully integrated into dozens of factories globally not just China, but Europe, southeast Asia, north America each logging thousands of hours a month without constant human oversight.

Lin:

I think the bigger ambition here, and maybe the story beneath that headline, isn't just operational efficiency. It's also about fundamentally changing how we see robots as co-workers, not just tools, by 2027, I'd love for the conversation to have moved from can robots do this To how can we best collaborate with them? That's the cultural shift we're really after. So, speaking of ambitious futures, let me ask you do you think society is actually ready for robots to become routine co-workers?

Luke:

transitions, in part, or maybe even mainly, because the robots are humanoid and this makes the perspective on being replaced much more personal, in a way. I'm expecting there to be significant resistance from organized labor, which will probably exacerbate the gap between those that do adjust and those that don't, and it's going to be very interesting to see how governments react and which side they support. I think, if you take the recent trillion yen National Robotics Fund that China just launched, that's there to speed adoption. I think that's what it's explicitly targeted at. So let's think about that organized labor side. What do you think that the unions Chinese, western, wherever they get right or they get wrong about the journey that we're actually on at the moment?

Lin:

That's such a critical point. Humanoids are fundamentally different psychologically because they look well human, and that makes the threat feel personal in a way, other automation never did. I think unions generally do understand one key thing correctly they see clearly that automation isn't just another wave. It's a transformation that will reshape the very definition of work. They're absolutely right to push hard for guarantees around retraining, upskilling and protecting the dignity of workers, around retraining, upskilling and protecting the dignity of workers. But what they sometimes get wrong, or at least underestimate, is the speed and inevitability of this shift.

Lin:

This isn't automation that's going to take decades, as you said. The next 30 months are going to feel incredibly fast. For example, china's 1 trillion yuan robotics fund you mentioned isn't just financial backing. It's a signal that this transition is now a national economic strategy. Where unions risk misstepping, in my view, is if they focus exclusively on preserving the past rather than shaping the future. Fighting automation itself is a losing battle, but unions have incredible leverage right now to insist on meaningful transition programs, things like funding for workers to move into roles such as robot supervisors or the DexOps engineers we're talking about. The key is proactive negotiation about what's next rather than a reactive defense of what's inevitably disappearing next, rather than a reactive defense of what's inevitably disappearing.

Luke:

So let's talk about the economics of robotics. What part of the robotics value chain do you think the profits are going to flow to in the future? Where do you see the margins being best in this?

Lin:

It's a great question. Hardware itself will commoditize fast. Just look at how quickly sensor costs have collapsed. The real margin is going to flow into software platforms and AI task orchestration the stuff that makes robots smart and easy to use. Think robotic software as a service, subscription models where factories pay continuously for task updates, fleet management and optimization. That's the future sweet spot.

Luke:

And any companies out there today that are really impressing you in that space.

Lin:

Definitely. Some standout players today are companies like Intrinsic, google's spin-out that's nailing AI-driven robot programming, covariant, which is doing incredible things with robot perception and pick-and-place software, vm Robotics, making it much easier to manage diverse robot fleets, and I'd also highlight Realtime Robotics, which is solving tough motion planning problems for complex manufacturing setups.

Luke:

Those are the ones I think are really onto something big right now right now and from a retail investor's perspective, who's worth keeping an eye on over the next three to five years to capture this expected growth in robotics.

Lin:

So I'm not giving financial advice, but if you want the under the radar picks, then let's talk about who's quietly building the infrastructure. One company I'd be watching closely is Symbotic. They're not exactly tiny, they're listed, but they're still early in their trajectory and they're doing really smart things with warehouse automation Not humanoids yet, but their modular systems are the kind of architecture that could adapt quickly once general-purpose robots become viable for logistics purpose robots become viable for logistics. Another one is Vicarious, surgical. They're in the surgical robotics space and while that feels like a niche, their work on minimally invasive manipulation overlaps in surprising ways with what we're trying to do in fine motor humanoid control.

Lin:

And I'd keep an eye on Sarko's technology and robotics. They've struggled a bit commercially on Sarko's technology and robotics. They've struggled a bit commercially, but their exoskeleton and teleop systems are a stepping stone toward collaborative humanoids in harsh or safety-critical environments. If they pivot or partner well, that upside could get real interesting, real fast. One more that's a little off the radar Keyence. They're huge in sensors and machine vision in Asia. Tactile and vision fusion is the next frontier in robot perception and they're sitting right on top of that. These aren't companies that are dominating headlines, but they're building parts of the puzzle and if humanoid adoption plays out the way, I think it will over the next three to five years. They could be holding key IP or integration layers when the big guys come calling.

Luke:

And if humanoid adoption plays out the way, I think it will over the next three to five years they could be holding key IP or integration layers when the big guys come calling and maybe taking a different perspective if you look at traditional businesses that can be modernized with robots over the next couple of years, which fields or trades or business types do you think have the greatest potential for being disrupted?

Lin:

That's such a sharp angle because, honestly, that's where a lot of the real value creation is going to happen Not just building the robots, but applying them in places that haven't caught up yet. If I were buying a traditional business today with the goal of modernizing it using humanoids, I'd be looking at light manufacturing and assembly, especially in sectors like furniture production, HVAC units or modular electrical systems. These are industries where the tasks are repetitive but just varied enough that traditional automation has struggled. Another space Commercial laundry operations. I know it's not glamorous, but folding, sorting, lifting, those are all high-repetition, physically demanding jobs that humanoids will crush as soon as fine motor dexterity improves just a bit more. Plus, there's already pressure around labor shortages in that space.

Lin:

And one more Construction materials prep Not full-on construction, but warehousing, cutting, packaging of things like pipe, cable, tile. It's heavy, it's inconsistent and it's often done in brutal environments. That's exactly the kind of niche where humanoids with tactile sensing and ruggedized limbs could shine Mine. Of course you'd need strong operational leadership and an integration partner that really understands robotics workflows, but the value unlock Massive.

Luke:

So I've heard Musk claim that a future Optimus robot could sell for $20,000 to $30,000 once it's been scaled, and given the speed of improvement, I'm sure that our CFO listeners are thinking, well, why not just wait for a year for prices to drop? So walk me through how you see robots as a service in manufacturing, making a material impact to companies' bottom line this year, in 2025, rather than in the future.

Lin:

Look, waiting can sound smart on paper, but in manufacturing, time is money real money. Robots as a service or RAS means no big upfront capex. Instead, manufacturers pay by the hour, around 85 yen per hour in our case, compared to roughly 140 yen per hour for triple-shift human labor. The math immediately tilts in favor of adopting now rather than later. Within weeks, factories can boost their throughput significantly, like at the automotive OEM, where we've already seen double-digit improvements in productivity just from consistent uptime and quick changeovers. Waiting a year doesn't just mean delaying investment. It means actively leaving money on the table every single day.

Luke:

Yeah, you know, I think that shifting implant automation to an as-a-service model is honestly going to be quite the cultural change, lynn. It will be interesting to see how companies take that up. Now, going back to what you said a moment ago, you mentioned this term, dexops engineer, which is a new one to me. What is that, and what does it mean relative to the people that are being replaced on these lines?

Lin:

Exactly right. A DexOps engineer is essentially someone who supervises and trains these humanoid robots using tactile and language feedback rather than traditional coding. It's less about manual labor and more about coaching robots to perform tasks effectively. So, yeah, many of those former line workers aren't disappearing. They're shifting into roles that combine their practical know-how with new technical skills.

Luke:

Not to pour too much cold water on this, but being realistic, Lynn, one DexOps engineer is going to end up managing many actual robots, right, and there's going to be a lot of people who don't make the transition that you're talking about. And there's going to be a lot of people who don't make the transition that you're talking about. Tell me maybe a bit more about how you expect humanoids to end up reshaping society over the let's say the next five years.

Lin:

Fair point, this transition isn't going to be comfortable or easy for everyone. Over the next five years, the shift to humanoids at scale means we're not just tweaking factories, we're restructuring whole job categories. One DexOps engineer might oversee 5, 10, maybe even 20 robots, which means significant job displacement is inevitable. We're talking potentially millions of manufacturing roles worldwide impacted, and that's no small thing. In practical terms, society faces two simultaneous challenges. First there's the economic disruption. People who've spent decades mastering a specific skill set assembling cars, packing boxes, doing quality inspections will suddenly find those skills far less valuable overnight Not everyone will, or realistically even can, smoothly transition into a high-tech supervisory role. Older workers especially may struggle to adapt fast enough to stay relevant, and we have to accept that many won't manage that leap without substantial support.

Lin:

The second, more subtle impact is psychological and cultural. Humans naturally build identity and dignity around their jobs. Now suddenly they're confronted with machines that don't just do the job better, they do it tirelessly, at lower cost and without complaint. That's tough to process and it might feed resentment, division, even political unrest, unless we proactively address it. Governments will need to step up their game dramatically, not just with symbolic retraining programs, but with real, effective and adequately funded initiatives. But let me be constructive about this. Two key things governments need to do immediately are first, invest in targeted retraining programs, not generic online courses, but hands-on, factory-level training programs that directly prepare displaced workers for roles like and other robotic management positions. And second, governments need to build and fund robust transitional income support, something like a wage insurance scheme, so people can realistically survive the transition period without collapsing financially or emotionally. Without these steps, we risk creating a two-tiered society those who successfully transition into this new tech economy and those left behind, and that, I think, would be the real tragedy here.

Luke:

Honestly, that's a sobering perspective and I'm absolutely with you that governments are going to need to take this seriously. I've got a couple of quickfire questions for you before we wrap up, okay. So first one which industry will humanoids quietly disrupt first, long before the media actually notices?

Lin:

Industrial agriculture. People will be shocked how quickly humanoids slip onto farms, sorting produce and managing harvests behind the scenes.

Luke:

Outside of China and the US, which country fields the largest fleet of robots by 2028, and why?

Lin:

Germany because their urgent demographic pressures, strong industrial base and serious public-private robotics investments create the perfect storm for rapid adoption.

Luke:

And when do we all get live-in robots like they have an iRobot?

Lin:

Honestly significant in-home humanoids like widespread adoption, I'd say late 2030s. Factories and warehouses have clear, immediate ROI cases, but homes present a completely different complexity in terms of unpredictability, safety and cost. We'll see limited deployments like specialized elderly care or assistive robots earlier, probably around the early 2030s, but the big wave robots folding laundry, cooking dinner and babysitting the kids is realistically about 15 years out.

Luke:

Thanks for the masterclass. This has been absolutely fascinating, although I was hoping for a personal laundry folding robot a little bit sooner than that. Now, where can our listeners dig deeper into the art of the possible with robots? What books, what channels, what do you recommend that they get into?

Lin:

Oh, absolutely. It's been fantastic talking through this. For anyone hungry to explore deeper, I'd highly recommend starting with Martin Ford's Rule of the Robots. It's a sharp, thoughtful look into how robotics and AI are reshaping our economy and society today. More recent but equally compelling is the Coming Wave by Mustafa Suleiman, which brilliantly covers how robotics and AI will reshape society in practical terms and AI will reshape society in practical terms. And if videos are your thing, check out the IEEE Spectrum Robotics channel on YouTube. It's an amazing place to see what's actually working in labs and factories right now.

Luke:

Zanedivis Lynn. Thank you again, and that's it for us today. Please like, subscribe and share if today's episode of AION sparked ideas. See you next time.