← BACK TO FEED
OpenAIroboticsembodied AIAGISam Altman

OpenAI Wants to Put a Robot in Every Home. First It Has to Build a Data Centre.

OpenAI is rebuilding its robotics division, which it shut down in 2020, with the near-term goal of using robots to help build infrastructure and the long-term vision of providing everyone with a personal general-purpose robot. The effort grew out of a world simulation research program led by Aditya Ramesh, and OpenAI is now hiring across hardware, operations, and machine learning. While Altman's ambitious end goal is likely years away, the underlying motivation may be generating valuable training data and advancing embodied AI models.

Sam Altman has gone public with a vision that sounds like a 1950s World's Fair exhibit: personal robots for everyone, capable of doing whatever you need. Whether that means fetching your dry cleaning or writing your quarterly report remains unspecified, which is doing a lot of work in that sentence.

For now, the ambitions are considerably more grounded. OpenAI is hiring across hardware, software, operations, and machine learning, with the near-term focus on robots that help specialists build out AI infrastructure. Think less domestic companion, more industrial site assistant. The gap between that and 'personal robot doing anything you need' is, to put it mildly, substantial.

The robotics effort apparently grew out of OpenAI's world simulation research programme, led by Aditya Ramesh. That group absorbed the Sora team after OpenAI quietly killed off its AI video app earlier this year, which is quite the career trajectory for a product that was being hyped as revolutionary not long ago.

This is also not OpenAI's first attempt at robotics. The company shut its original robotics division in 2020, with the stated rationale being that AGI could be reached faster without the messiness of physical hardware, and that usable robot training data was too thin on the ground. Both of those things remain partially true, which makes the revival interesting.

The rebuild started in January 2025, with the explicit goal of developing general-purpose robots as a path toward AGI. What that actually means in practice is vague. OpenAI recently announced a pivot toward AI agent applications, so where robots fit into that strategic picture is not entirely obvious.

The cynical read, and arguably the more plausible one, is that this is really about data. Embodied AI systems generate a category of training signal you simply cannot get from text or video. Real-world physical interaction produces exactly the kind of messy, grounded experience that current models notoriously lack. If OpenAI can build robots that operate in the real world at scale, the data flywheel that creates might be worth more than any individual product.

Altman's everyone-gets-a-robot future is, by any realistic estimate, a long way off. But the infrastructure play in the near term, and the embodied AI research angle underneath all of it, makes the investment less eccentric than the headline suggests.