 |
|
|
|||||
With one sweeping gesture, Dar Sleeper hoists the humanoid robot off the ground. Bracing its back with one arm and its legs with the other, he gently carries it across the room and lowers it onto a sofa, where it lies in repose as if catching a quick nap. It’s a slightly surreal scene, but it has a serious point. I am visiting the Palo Alto headquarters of 1X Technologies, and Sleeper, the company’s VP of growth, is demonstrating that Neo, its home robot, is a lightweight at a mere 66 pounds. That’s a crucial design feature, given that a weighty domestic bot could prove hazardous if it toppled over in the vicinity of a human, a pet, or just a pricey vase. Soon, Neo will take on the ultimate proving ground for a home robot: actual homes. 1X is announcing that its taking preorders and plans to ship units to its earliest customers next year. The price is $20,000, or $499 per month as a subscription service, with a six-month minimum. Like a smartphone, the robot will come in multiple color optionstan, gray, and dark brown. Wait, $20,000? There isnt much precedent for mainstream consumer products in that price range. Cars, of course. (The average price for a new one just topped $50,000.) Maybe boats? Even if you can come up with another example or two, its a short list. Then again, 1X founder and CEO Bernd Brnichs goals for his robots involve attaining a degree of utility that few inventions ever have. He wants to teach Neo to handle every household task that people perform because they need to, not because they want to. Even if the time saved came in bits and piecesfive minutes of dishwasher unloading here, 15 minutes of laundry folding thereit would add up to many hours newly available for more rewarding pursuits. [Photo: Courtesy 1X Technologies] Now multiply that by hundreds of thousands of Neos, which is the quantity Brnich talks about when outlining 1Xs long-term plans for manufacturing capacity. Potentially it’s the biggest thing ever, with respect to productivity increase across society, he tells me when I catch up with him via Zoom after my visit to the company. Brnich is hardly the only entrepreneur spinning visions of a world in which humans coexist with teeming masses of robots modeled on them. Companies such as Agility Robotics, Apptronik, and Figure AI are all building humanoid bots of their own. So is Sleepers former employer, Tesla, whose Optimus robot has seemingly diverted much of Elon Musks attention away from boring old EVs. Most of the nascent categorys focus is on environments such as factories and warehouses, where robots might take on jobs that are dangerous or repetitiveno paychecks required. (1X is presently focused on the home but says it plans to offer humanoids for the workplace at a later date.) Its not just robotics manufacturers that see a vast business in the making. Humanoid bots could fuel enormous growth in chips, foundational models, cloud services, and other elements of the AI economya wave of optimism 1X is already riding. In 2023, OpenAI led the companys $23.5 million Series A2 funding round; its Series B round the following year raised another $100 million. The company also has a strategic alliance with Nvidia, whose CEO, Jensen Huang, accepted a custom leather jacket from Neo during an onstage appearance last March. By 2050, according to a May Morgan Stanley report, there could be more than a billion humanoid robots in service, with the vast majority in commercial and industrial settings. Only 80 million would be in homes, the report speculates. Still, thats roughly 80 million more humanoid home robots than exist today. [Photo: Courtesy 1X Technologies] For all the grandeur of 1Xs plan to transform how we spend our time, the company is up front that Neo is not yet ready to handle all the drudgery you might want to offload to a home robot. Some of the things it can do are only possible under the control of remote 1X employees, who can orchestrate its actions as if it were a marionette. That human assistance might be a stopgap, but it will also provide 1X with AI training data it can use to make Neo more autonomous over time. My encounter with Neo at 1Xs office did as much to tamp down my impressions of its present form as ratchet them up. The robot gamely ambled about, with a gait that didnt closely resemble that of a human but was more limber than you might expect. It showed off its strength by hefting a restaurant-sized sack of rice. It also presented me with a chilled bottle of water, though the feat involved a mix of autonomy (walking to the refrigerator) and remote control (oening the fridge and fishing out the bottle), and took much longer than if I’d simply grabbed the water myself. A demo of Neos sweeping skills went south when it turned out the cordless vacuum it was wielding had a dead battery. (A human house cleaner would have easily diagnosed and solved the problem.) And some uses 1X is citing for its robotsuch as setting birthday reminders and managing grocery listsdo not seem to call for $20,000 of advanced machinery. For Neo to succeed, 1X wont just need to finish building out the robots skill set. It will also have to live up to expectations long steeped in cultural touchstones such as The Jetsons and Star Wars. As Sleeper puts it, “The ultimate North Star, in a lot of people’s minds, is Rosie the robot.” Brnich emphasizes that 1X isnt entirely sure how Neo will fit into homes, and the company is getting ready to sell them in part to help figure that out. I don’t think there’s any way we could be as creative as the full set of our customers, he says. That’s one of the beauties of actually developing this in public. Robots come home Originally known as Halodi Robotics and founded by Brnich in Norway in 2014, 1X didnt set out to build a home humanoid right away. Its first robot, Eve, was designed for industrial applications, had grippers instead of hands, and rolled on two wheels. It looked a little like it was riding a built-in Segway. As with anything dependent on generative AI, teaching robots to engage with the world requires vast amounts of training data. As 1X considered where to get that data, it came to see the home as rife with possibilities. That fact nudged it in the direction that eventually led to Neo. A cup has the ability to give you more data than an entire industrial task, says Sleeper. Because that cup could be full. It could be empty. It could be half full. It could be on the kitchen table full, meaning you should leave it there because somebody’s probably drinking it. It can be empty on the coffee table, which means you should take it to the sink or put it in the dishwasher. It could be in the cupboard, meaning that that’s where you go to grab it to give it to somebody when they ask it for it. Along with data, homes were also full of challenges that Eve had never been designed to overcome, such as climbing stairs to fetch a laundry basket. So when the company formally decided to pursue the home market, it realized it would need to build an all-new robot. Every lesson we’ve learned has moved us more and more in the direction of being very confident that the human form factor is the only form factor, says Sleeper. Humanoids were hardly a new field of robotics research: Brnich says he drew inspiration from examples such as Hondas Asimo, a staple of trade-show demos early in this century. But he adds that The type of hardware that we’ve designed here, which is so lightweight and safe and capable, we couldn’t have done that 20 years ago. Visiting 1Xs Palo Alto operation is certainly an evocative experience. Along with standard tech-company accoutrements such as HP 3-D printers cranking out prototype parts, its full of robotsfrom the Neo I watched in action to a row of decommissioned Eves sitting next to a staircase. One skinless Neo hangs from a hook in an area thrumming with human technicians: “It’s very Westworld in here,” says Sleeper. Much of Neos innovation lies underneath its knit turtleneck jumpsuit. 1X created its own motors, which it says have five times more torque than any other in the world and help the robot lift up to 154 pounds and carry up to 50 pounds. It equipped the robot with a tendon-drive transmission system, intended to reduce weight while increasing safety and affordability. It engineered hands with 22 degrees of freedomonly five less than human onesand attached them to longer-than-human arms, since robots have no shoulder blades. The company even custom-designed its own batteries. Neo officially runs for four hours on a charge, though it can top off its own battery as necessaryYoull never really have to think about it, promises Sleeper. [Photo: Courtesy 1X Technologies] At 5 feet 6 inches, Neo is 2.5 inches taller than the average American woman and 3 inches shorter than the average man. Its physique is intentionally neither feminine nor masculine: “It was important that nobody be sexually attracted to it,” says Sleeper, suggesting scenarios I prefer not to contemplate any further. Owners will, however, be able to choose between female and male speaking voices. (For the record, Sleeper referred to Neo as “he” during our conversation, but said that opinions on its gender, if any, vary.) Regardless of how capable Neos hardware may be, 1Xs AI platform will determine what it can actually do around a house. Its elements include machine vision, audio recognition, and an onboard LLM with memory that provides ongoing context. Melding all of them will allow the robot to listen for instructions, navigate its surroundings and the objects therein, and perform useful tasks. The LLM will also power Neos personality. That will come out as people talk to it, which 1X thinks they may do quite a lot. Rather than mimicking human companionship, the company is aiming to make the robot a lovable, pet-like sidekickHobbes to the owners Calvin, Sleeper says. All of this is where the experience that 1X plans to ship next year will be, at best, a first rough draft. The companys examples of jobs Neo will be able to perform autonomously on day one are modest: fetching items, greeting guests at the door, switching off the lights at night. It will handle these and other responsibilities by breaking them down into 10-second microtasks. At the moment, things get tricky if much more than three microtasks are involved. 1X expects its first customers to understand that theyre buying into a long-term proposition. As you continue to use your robot, you’ll get more updates, and your basic autonomy features will expand to be able to do more, says Sleeper. Your robot can always try something if you ask it to, like getting you a beer. And if it hasn’t learned how to do that yet efficiently, it will fail. Some people might hate it. But those arent the early users. (Historical footnote: Delivering beer may be home robotics most enduring example of an exciting application. Its what a bot called Androbot did for Atari cofounder Nolan Bushnell during an onstage demo at CES in 1983, to cheers from the audience.) From the beginning, Neo will be able to go beyond basic autonomy through its remote-assistance feature. Called Chores, it will enable California-based 1X employees to see through Neos eyes and steer the robot through more elaborate processes. The company has anticipated privacy concerns over its bot becoming a roaming window into its customers homes: Chores sessions are only initiated if users explicitly ask for them, and can be terminated at any time. When theyre active, the LED ring around Neos ears will change from white to blue. Remote-controlled robots made me think of the Tesla event held a year ago whose Optimus bartenders provoked amazement, at least at first. When it turned out theyd been puppeteered by off-site humans, the awe dissipated. Sleeper told me that 1X doesnt consider Chores to violate the fundamental idea of a home robot: We just think that there’s actually quite a lot of value to come from these expert operators being able to supplement and support your experience. As they do, the training data they generate may redound to the benefit of future Neo owners. To err is humanoid Brnich, who has been living with Neo at home for quite some time, professes not to be overly concerned with any flaws in the current experience, cheerfully calling them robotic slop. Maybe Neo didn’t put all of the glasses in the cabinet exactly in line, like this master Japanese butler, but it put them in there, he says by way of example. Okay, the shirt isn’t folded perfectly, but it’s folded pretty well, and I didn’t need to do it. I’m actually very happy. But he also expects that Neo will quickly prove its worth to a fairly broad audience. How soon? By 2027, the purchase will be a no-brainer, he predicts. Comparing home robots to smartphoneswhich have tended to get more powerful over time rather than much cheaperBrnich says that $20,000 may remain Neos price tag even after 1X can scale up production well beyond its current capacity of 25,000 U.S.-made robots a year. You just want to keep expanding the capabilities until you can not only do what humans can do, but you can also do a lot of things that we can’t, he explains. A lot has to go right for these ambitions to play out, including 1X having access to the necessary capital. Last month, The Informations Wayne Ma, Natasha Mascarenhas, and Valida Pau reported that the company was seeking $1 billion in additional funding at a $10 billion valuation, dwarfing its previous numbers. For now, 1X is still working on teaching Neo to tackle work that humans can literally do with their eyes closed. But when I ask Brnich if he considers the companys ultimate business model to lie in selling home robots or building services around them, he pushes back on the assumption that its a home robotics company at all. Actually, I think of 1X mainly as an AGI company, he says. We want to solve intelligence and solve artificial labor across society. Digitally and physicallythe entire substrate. And you’re not going get there without robots. Now 1X has to prove that it will get there with them.
Category:
E-Commerce
OpenAI said Tuesday it has reorganized its ownership structure and converted its business into a public benefit corporation and two crucial regulators, the Delaware and California attorneys general, said they would not oppose the plan. The restructuring paves the way for the ChatGPT maker to more easily profit off its artificial intelligence technology even as it remains technically under the control of a nonprofit. Delaware Attorney General Kathy Jennings and California Attorney General Rob Bonta said in separate statements that they would not object to the proposal, seemingly bringing to an end more than a year of negotiations and announcements about the future of OpenAI’s governance and the power that for-profit investors and its nonprofit board will have over the organization’s technology. The company also said it has signed a new agreement with its longtime backer Microsoft that gives the software giant a roughly 27% stake in OpenAI’s new for-profit corporation but changes some of the details of their close partnership. The attorneys general of Delaware, where OpenAI is incorporated, and California, where it is headquartered, had both said theyre investigating the proposed changes. We will be keeping a close eye on OpenAI to ensure ongoing adherence to its charitable mission and the protection of the safety of all Californians, said Bonta. OpenAI said it completed its restructuring after nearly a year of engaging in constructive dialogue with the offices in both states. OpenAI has completed its recapitalization, simplifying its corporate structure, said a blog post Tuesday from Bret Taylor, the chair of OpenAI’s board of directors. The nonprofit remains in control of the for-profit, and now has a direct path to major resources before AGI arrives. AGI stands for artificial general intelligence, which OpenAI defines as highly autonomous systems that outperform humans at most economically valuable work. OpenAI was founded as a nonprofit in 2015 with a mission to safely build AGI for humanity’s benefit. It later started a for-profit arm. Microsoft invested its first $1 billion in OpenAI in 2019 and the two companies formed an agreement that made Microsoft the exclusive provider of the computing power needed to build OpenAIs technology. In turn, Microsoft heavily used the technology behind ChatGPT to enhance its own AI products. The two companies first revealed in January that they were altering that agreement, enabling San Francisco-based OpenAI to build its own computing capacity, primarily for research and training of models. That coincided with OpenAIs announcements of a partnership with Oracle and SoftBank to build a massive new data center in Abilene, Texas. It’s since announced more such projects planned in the U.S., Asia, Europe and South America, along with big deals with chipmakers like Nvidia, AMD and Broadcom. But other parts of its agreements with Microsoft remained up in the air as the two companies appeared to veer further apart before reaching a tentative new agreement in September. OpenAI had previously said its own nonprofit board will decide when AGI is reached, effectively ending its Microsoft partnership. But it now says that once AGI is declared by OpenAI, that declaration will now be verified by an independent expert panel, and that Microsoft’s rights to OpenAI’s confidential research methods will remain until either the expert panel verifies AGI or through 2030, whichever is first. Microsoft will also retain some commercial rights to OpenAI products post-AGI and through 2032. Microsoft put out the same joint announcement about the revised partnership Tuesday but declined further comment. Its shares spiked 2% on Tuesday. Going forward, the nonprofit will be called the OpenAI Foundation and Taylor said it would grant out $25 billion toward health and curing diseases and protecting against the cybersecurity risks of AI. He did not say over what time period those funds would be dispersed. Robert Weissman, co-president of the nonprofit Public Citizen, said this arrangement does not guarantee the nonprofit independence, likening it to a corporate foundation that will serve the interests of the for profit. Even as the nonprofits board may technically remain in control, Weissman said that control is illusory because there is no evidence of the nonprofit ever imposing its values on the for profit. The Delaware attorney generals investigation focused on ensuring OpenAI put its commitment to safety first and before any financial interests. Jennings also said OpenAI promised to keep its nonprofit in control of the public benefit corporation, including the right to appoint and remove its board members. The removal of OpenAI’s CEO Sam Altman in Nov. 2023 by the nonprofit’s board at the time and his subsequent reappointment kicked off the company’s effort to restructure. The nonprofits board will continue to include a Safety and Security Committee, which will have the power to oversee and review OpenAIs technology development. It will even have the power to stop the release of a new product, according to the Delaware attorney generals statement. Additionally, within a year, the nonprofits board will include at least two members who do not also serve on the public benefit corporations board. OpenAI still faces a legal challenge from billionaire Tesla CEO Elon Musk, an early OpenAI investor who now runs his own AI firm, xAI, and has accused the startup he co-founded of betraying its original mission. A federal judge in March denied Musks request for a court order blocking OpenAI from converting itself to a for-profit company but said she could expedite a trial to consider Musks claims. Matt O’Brien and Thalia Beaty
Category:
E-Commerce
In artificial intelligence, compute and data matter, but people matter more. Behind every breakthrough model, every infrastructure leap, and every revolutionary chatbot lies a shrinking pool of scientists, engineers, and mathematicians capable of building them. The defining constraint on the next decade of AI isnt just hardware: its human capital. Across the world, a quiet arms race is unfolding for that capital. The most advanced AI firms, like OpenAI, Anthropic, DeepMind, Meta, Google, and a few in China, are no longer competing just for customers or GPUs. They are competing for brains. The new concentration of intelligence In the past two years, the hiring and acquisition patterns of AI companies have begun to resemble a geopolitical map. Anthropic and OpenAI lure entire research teams from Google or Meta with compensation packages approaching nine figures. Apple and Amazon, late to the party, are buying startups not for products, but for the engineers behind them. And venture capital is no longer funding ideas so much as acqui-hiring: purchasing human potential before it matures elsewhere. Multiple analyses show that elite U.S. programs, especially Stanford, Berkeley, Carnegie Mellon, and MIT remain dominant feeders into frontier AI labs, reinforcing a tight concentration of expertise in a few firms and geographies. The result is intellectual concentration unprecedented in the history of technology. This clustering may accelerate progress in the short term. But it also increases fragility. When innovation lives inside a handful of firms, the industry becomes monocultural. The same assumptions, ethical frameworks, and commercial incentives repeat themselves. Alternative approaches, like symbolic reasoning, hybrid models, and decentralized architectures, struggle for attention or funding. The global scramble for minds Meanwhile, countries are treating AI researchers the way they once treated nuclear physicists or oil engineers. The United Kingdom launched its Frontier AI Taskforce with special visas for top scientists. Canadas Global Talent Stream fast-tracks work permits for AI engineers in under two weeks. France offers tax incentives and research grants for companies that locate their labs in Paris or Grenoble. China, faced with export controls on chips, has doubled down on human intelligence as a strategic resource. Its leading universities are producing tens of thousands of AI graduates a year, many trained on open-weight models after Washingtons hardware restrictions. As one analyst at the Carnegie Endowment put it, If you cant import compute, you import talent. In other words: brains are the new semiconductors. Americas self-inflicted wound And yet, the United States, still home to most of the worlds top AI firms, is busy tightening the spigot. Donald Trumps renewed hostility toward immigration includes threats to limit, suspend, or impose heavy fees on H-1B and F-1 visas, the very programs through which thousands of AI researchers enter the country each year. The pattern is not new. In 2020, during his previous term, Trump signed an executive order suspending key visa categories, prompting MIT Technology Review to warn that the move threatened to undercut Americas lead in artificial intelligence. The logic hasnt changed. The worlds best AI researchers are disproportionately international: roughly 60% of top AI scientists working in the U.S. were born abroad, according to the National Foundation for American Policy. Limiting their entry isnt protectionism: its strategic sabotage. When your biggest competitive advantage is talent, closing the door to talent is a slow form of self-destruction. The corporate paradox Ironically, the companies that benefit from global talent the most are also the ones making that talent scarce. By offering astronomical salaries and exclusivity contracts, they create a gravitational field that pulls expertise out of universities and startups. Academia, long the seedbed of AI progress, is now bleeding researchers to industry at an unprecedented rate. The result is a research vacuum where public institutions can no longer afford to compete. Even national labs struggle to retain talent when private firms offer multiples of government pay. This corporate concentration of minds has another cost: intellectual homogeneity. When the same people cycle between the same companies under the same investors, the frontier of AI becomes narrower, more predictable, and less plural. The next big breakthroughs might never happen, and not because we lack compute, but because weve trained the global research community to think the same way. The geopolitical stakes For decades, the United States has dominated global innovation by acting as a magnet for talent. The H-1B and F-1 visa systems, for all their flaws, turned American universities and tech hubs into engines of discovery. That advantage is now at risk. If Washington continues down the path of visa restriction, and if the industry continues to hoard rather than nurture expertise, the gravitational center of AI could shift elsewhere. Canada, the European Union, and the UAE are already competing for displaced researchers. China, meanwhile, is developing homegrown alternatives with staggering speed. The irony is that America may lose the very thing that made its technology ecosystem unstoppable: openness. The more insular it becomes, the more it resembles the centralized systems it once out-innovated. The ethics of scarcity Talent concentration also raises moral questions. When a small number of corporations control the majority of global AI expertise, they effectively control which problems get solved and which are ignored. We are already seeing this bias in practice. Billions are being poured into models that optimize productivity, marketing, and financial forecasting, while underfunded projects in climate modeling, education, and healthcare languish. AIs promise to benefit humanity is hollow if humanity doesnt have a seat at the table. Diversity of thought, background, and geography is not a moral luxury; its a prerequisite for resilience. Homogeneous systems fail in homogeneous ways. A new social contract for intelligence The solution is neither regulation alone nor market freedom alone. Its a new social contract for talentone that treats human intelligence as a shared strategic resource, not a proprietary asset. That means: Immigration policies that attract, not repel, the worlds brightest minds. Funding mechanisms that keep public research competitive with corporate labs. Ethical frameworks that prevent non-competes and exclusivity contracts from turning scientists into captives. Global cooperation that recognizes AI as a common infrastructure challenge, not a zero-sum race. In the 20th century, nations competed for oil. In the 21st, they will compete for cognition. A warning and a choice The United States still holds the advantage: world-class universities, deep capital markets, and a culture of risk. But advantages erode when arrogance replaces openness. If America turns inward, it wont just lose talent: it will lose the diversity that fuels creativity. And when the next generation of scientists chooses to work in Toronto, Paris, Madrid, or Shenzhen instead of Silicon Valley, the American century of innovation will quietly end. Not with a revolution, but with a resignation letter. Computing power, as I said in previous articles, is important, as it is access to cheap energy, or as Europe very well knows, regulation. But the race for AI supremacy will not be won by compute, nor by energy prices or by policy alone. It will be won by whoever attracts and empowers the minds that make intelligence itself. And right now, those minds are watching which countries still deserve them.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||