OpenAI is building robots — but Tesla has a massive head start
OpenAI officially launched a robotics division on May 31, 2026, signaling that the company best known for ChatGPT now wants a physical presence in the world. The team grew out of an internal world-simulation research program led by Aditya Ramesh. With humanoid robots projected to become a $5 trillion market by 2050, per Morgan Stanley, the stakes are real — even if OpenAI's timeline is not yet.
The plan
OpenAI's strategy runs in two phases. Near-term, the team will focus on robots that assist skilled workers in construction and infrastructure — environments too unpredictable for conventional industrial arms. The longer-term goal is a personal robot for everyday home use, a direct challenge to Tesla, Boston Dynamics, and a crowded field of humanoid startups.
The defining technical bet is hardware-software co-design: rather than building a mechanical body first and bolting on AI afterward, OpenAI says the hardware and the machine-learning models will be developed together. The idea is to avoid the constraints that off-the-shelf robotics hardware typically imposes on what an AI can actually do. The company is actively hiring hardware engineers, systems architects, and ML specialists — applications go to robotics-recruiting@openai.com.
OpenAI Robotics is hiring, looking for exceptional full-stack hardware, ops, systems, and ML engineers to help us program and manufacture robots that are useful for society.
— Sam Altman (@sama) May 31, 2026
AI should be able to help people in the physical world. In the short term, we are focused on robots to…
This is OpenAI's second attempt at robotics. The company shut its first division down in 2021, citing the difficulty of training algorithms for physical tasks. The relaunch reflects confidence in newer neural-network architectures — but also raises questions. Caitlin Kalinowski, OpenAI's hardware lead, resigned in March 2026 over concerns about the company's Pentagon deal and surveillance applications. Her departure leaves a visible gap in the team's leadership.
Tesla's concrete roadmap
While OpenAI's path remains opaque, Tesla has numbers on the board. The company plans to produce 5,000–10,000 Optimus units in 2025 for internal factory use, scaling to roughly one million units annually by end-2026, according to Tesla Optimus Timeline reporting. At that volume, the per-unit cost is estimated to fall below $20,000 — potentially competitive with industrial automation equipment.
OpenAI's former robotics partner, Figure AI, ended its collaboration in February 2025 to develop proprietary AI in-house. That leaves OpenAI without a clear hardware ally as it restarts — a contrast to Tesla's vertically integrated approach at Giga Texas.
A race with unequal starting lines
Expert opinion is split on whether humanoid robots will deliver at scale. Roboticists including Rodney Brooks argue specialized, task-specific robots will outperform general-purpose humanoids for years. OpenAI is betting otherwise, but without a disclosed manufacturing plan or hardware partner, the robotics division is still far more promise than product. The OpenAI Robotics Announcement is a signal of intent — the hardware is another story entirely.