{"slug": "apptronik-is-turning-robot-training-into-the-next-ai-factory", "title": "Apptronik is turning robot training into the next AI factory", "summary": "Apptronik opened a 90,000-square-foot Robot Park in Austin where its Apollo humanoid robots practice warehouse tasks, with most sessions still guided by human operators. The facility, described as a data factory, generates training data for the robots, which are being tested by Mercedes-Benz for logistics tasks. Apptronik has raised about $1 billion and is valued at over $5.5 billion, with Google DeepMind adapting its Gemini Robotics model to Apollo.", "body_md": "*Apptronik's new Austin Robot Park is less a showroom than a working data floor. If humanoid robots are going to earn their keep, this is the kind of unglamorous training infrastructure they need first.*\n\nApptronik has opened a nearly 90,000-square-foot Robot Park in Austin where its Apollo humanoids practice loading boxes, sorting objects and doing the dull physical work that looks easy only until a robot has to do it without falling over. According to Business Insider, which reported the launch on June 30, most of those sessions are still guided by human operators standing nearby or controlling the machines remotely.\n\nThat detail is the story. You can watch enough humanoid robot videos online to believe the hard part is already over. It isn't. The hard part is turning a few clean demonstrations into repeatable work at customer sites, with safety rules, damaged packaging, odd lighting, tired workers, and a robot that has to learn from more than a perfect lab table.\n\nApptronik cofounder and CEO Jeff Cardenas described the facility to Business Insider as a data factory, and that's the right phrase. Chatbots had the internet to read. Robots don't have a giant public archive of hands loading totes, arms reaching awkwardly across a conveyor, or feet adjusting around a cluttered warehouse floor. Someone has to manufacture that experience. Apptronik is betting it can do that before the economics of humanoids harden around someone else.\n\nThe Austin company has raised about $1 billion and is valued at more than $5.5 billion, according to Business Insider. Earlier reporting from Axios said Apptronik raised $350 million in Series A funding in February 2025, co-led by B Capital and Capital Factory, with Google participating. El Pais later reported that the Series A had reached $935 million and valued the company at about $5.3 billion, with backers including Alphabet, Mercedes-Benz, Peak6, B Capital, AT&T Ventures, John Deere and Qatar Investment Authority.\n\nThose numbers buy time, but they don't buy deployment. Apollo 2, the current version, is built for data collection and customer pilots, with upgraded batteries, motors and sensors. Business Insider reported that it stands around 6 feet tall, runs for four hours and can lift 55 pounds with both hands. Apollo 3 is supposed to be the commercial workhorse, but Cardenas didn't give Business Insider a launch date. Frankly, that silence is more useful than another heroic timeline. If you sell the date before the robot can survive the job, the customer becomes your test rig.\n\nMercedes is the most concrete customer signal here. Axios reported in 2024 that Mercedes-Benz was exploring Apollo for logistics tasks, including bringing parts to assembly workers, inspecting components and delivering totes of kitted parts. Business Insider now says Mercedes uses Apollo robots in its factories for simple tasks such as gathering components and tools for assembly-line jobs. That isn't science fiction. It's also not a finished labor replacement. It's the messy middle where a machine starts by doing narrow, repetitive work and earns trust one shift at a time.\n\n## Google gives the brain a body\n\nApptronik's Google DeepMind relationship matters because humanoid robotics is no longer only a hardware contest. The Verge reported in June 2025 that Google DeepMind adapted its on-device Gemini Robotics model to Apptronik's Apollo, after training the model on Google's ALOHA robot, and said the model could adapt to new situations with as few as 50 to 100 demonstrations. That doesn't make Apollo autonomous in the way the public imagines. It means the software side is beginning to care about different robot bodies, not just one perfectly controlled machine.\n\nYou should read Robot Park through that lens. Human-guided sessions are not an embarrassment. They are the raw material. The operator shows the motion, the robot records the physical context, and the model slowly gets less useless at the edge cases that make warehouses hard. 1X is running into a similar truth in homes. The New Yorker reported this week that its Neo robot still relies heavily on tele-operators and struggles with autonomy in real-world settings. The setting changes, but the bottleneck is familiar: physical AI needs examples from the physical world.\n\nThe competitive field is getting crowded fast. Business Insider reported that Figure AI was most recently valued at $39 billion and is moving into logistics and distribution center deployments. Agility Robotics is further along in one obvious way: Business Insider and The Wall Street Journal reported last week that Agility plans to go public through a SPAC deal valuing it at $2.5 billion, with Digit already used by customers including Amazon, Toyota, GXO and Schaeffler. Tesla's Optimus push still looms over the whole category because Tesla has factories, cameras, compute and Elon Musk's talent for forcing attention onto a product before the market has sorted out what is real.\n\nApptronik's answer is not to win the video contest. It is to build the place where robots learn boring work. The company also has both legged and wheeled versions of Apollo, according to Business Insider, with Cardenas saying wheeled robots may deploy sooner because they are safer while legged systems carry more long-term promise. That is a practical distinction. A robot that can walk upstairs is impressive. A robot that doesn't fall into a human coworker may be worth more on the first purchase order.\n\nFor startups watching this market, the useful lesson is simple: embodied AI won't scale on model releases alone. It needs facilities, operators, customer pilots, safety procedures, service teams and enough task data to make autonomy cheaper than remote control. Apptronik's Robot Park is current because it shows where the humanoid race has really moved. Not from lab video to mass adoption, but from lab video to training floor.\n\nThat's still progress. Just don't confuse school with graduation.\n\n**Also read:** [The Bank of England is warning that AI trading agents could trigger the next market crisis](https://startupfortune.com/the-bank-of-england-is-warning-that-ai-trading-agents-could-trigger-the-next-market-crisis/) • [Call center giants are being repriced out of existence before AI has finished the job](https://startupfortune.com/call-center-giants-are-being-repriced-out-of-existence-before-ai-has-finished-the-job/) • [The AI notetaker sitting in your Zoom call may be your next legal liability](https://startupfortune.com/the-ai-notetaker-sitting-in-your-zoom-call-may-be-your-next-legal-liability/)", "url": "https://wpnews.pro/news/apptronik-is-turning-robot-training-into-the-next-ai-factory", "canonical_source": "https://startupfortune.com/apptronik-is-turning-robot-training-into-the-next-ai-factory/", "published_at": "2026-06-30 12:59:33+00:00", "updated_at": "2026-06-30 13:30:51.490778+00:00", "lang": "en", "topics": ["robotics", "artificial-intelligence", "ai-infrastructure", "ai-startups", "ai-products"], "entities": ["Apptronik", "Apollo", "Google DeepMind", "Mercedes-Benz", "B Capital", "Capital Factory", "Alphabet", "Jeff Cardenas"], "alternates": {"html": "https://wpnews.pro/news/apptronik-is-turning-robot-training-into-the-next-ai-factory", "markdown": "https://wpnews.pro/news/apptronik-is-turning-robot-training-into-the-next-ai-factory.md", "text": "https://wpnews.pro/news/apptronik-is-turning-robot-training-into-the-next-ai-factory.txt", "jsonld": "https://wpnews.pro/news/apptronik-is-turning-robot-training-into-the-next-ai-factory.jsonld"}}