OpenAI Officially Establishes Robotics Team, Focusing on Construction Workers Assistance
Miles Bennett
OpenAI has upgraded robotics from a portfolio bet to an in-house strategic unit, targeting construction sites rather than living rooms. This means a foundation-model company is now building its own hardware — reshaping who competes with whom.
Why is OpenAI suddenly building robots itself?
On June 1, co-founder Greg Brockman announced the "OpenAI Robotics" team is ramping fast, with open roles for full-stack hardware, operations, systems, and ML engineers.
Until now OpenAI only held minority stakes in humanoid-robotics firms Figure and 1X — no in-house hardware. This means → robotics has moved from "write a check and watch" to "sit at the table and play."
In plain terms = OpenAI was an investor; now it is a builder.
Why construction sites first — not home robots?
CEO Sam Altman said the same day: the near-term focus is assisting skilled tradespeople in construction and physical infrastructure, not consumer or household robots.
Altman's long-term vision is "a personal robot for everyone," but no product, partner, or timeline has been disclosed.
This reflects a deliberate choice of a well-defined, controllable entry point — success is easier to measure on a job site than in a kitchen.
Why did the spin-off plan fall through?
The Wall Street Journal reported that Altman discussed spinning out the robotics and consumer-hardware units for independent fundraising late last year.
The plan stalled because the spun-out entities would likely still consolidate onto the parent's balance sheet. Insiders described the two units as "standalone startups sheltered under OpenAI's roof."
Put simply = even after a split, the parent's books wouldn't get any lighter — so the split wasn't worth doing.
What signal does the timing send?
The announcement lands just after OpenAI closed a $122 billion funding round at an $852 billion valuation, with an IPO expected this year.
Yet per the Wall Street Journal, revenue and user growth have both missed internal targets. CFO Sarah Friar warned that sluggish revenue growth could impair the company's ability to honor data-center commitments.
This reflects a need for a fresh narrative ahead of the IPO — when the core business isn't growing fast enough, a bold new storyline helps prop up the valuation.
What does this mean for the robot startups OpenAI already backs?
Google DeepMind, Tesla, and Boston Dynamics have all deployed scaled physical-AI systems. OpenAI is a late entrant.
Its edge is foundation-model capability, but whether that translates into a robotics product remains unproven.
This means → foundation-model intelligence may be embedded directly into OpenAI's own robot platform rather than sold as an external API. For portfolio companies like Figure and 1X, the partnership and the competition just deepened at the same time.
We are trying to figure out how to actually succeed in robotics.
Sam Altman
OpenAI CEO
(Core Memory podcast interview)
Content is for reference only, not financial advice.