OpenAI CEO Sam Altman held discussions late last year about plans to spin off the company's robotics division and consumer hardware unit into independently operated entities. The move was intended to grant both businesses greater development flexibility while preventing them from weighing down the group's core primary operations.
According to informed sources, the initial proposal would have allowed the two newly separated companies to raise external funding independently and operate autonomously. However, this plan was ultimately rejected, partly because OpenAI determined after assessment that these new entities would still need to be consolidated into the group's balance sheet, making full financial isolation unachievable.
This spin-off proposal highlights the difficult trade-offs OpenAI faces as it pushes toward an IPO. For years, Altman has greenlit numerous ambitious new projects that extend far beyond the popular ChatGPT, aiming to set a high bar for the company's scientific research and product development. However, the company is now under increasing pressure to rein in control over "side projects" that do not substantially contribute to revenue growth.
Following its position behind competitor Anthropic, OpenAI is recalibrating its business focus, concentrating efforts on building a new super-app to attract more developers and enterprise users. The company recently missed some internal targets for user growth and revenue, though officials stated that business growth has accelerated following the release of new AI models. Concurrently, OpenAI has discontinued its video generation tool Sora to free up computational resources for core products.
Insiders suggest that OpenAI might revisit the idea of spinning off business lines in the future. A potential model could emulate the structure Alphabet established in 2015: separating Google's core search business from higher-risk, cutting-edge innovation units like the self-driving car venture Waymo and life sciences unit Verily.
In its financial reports, Alphabet separately presents revenue from cash cows like Google Search, YouTube, and cloud computing, alongside the operational losses of its frontier innovation businesses. This structure allows investors to separately evaluate the performance of Google's main operations and the scale of the group's long-term strategic investments.
Other tech giants, while not adopting the same holding company structure, provide similar financial transparency: Meta separately discloses the multi-billion dollar annual losses of its Metaverse division, and Microsoft breaks out the operating performance of segments like its gaming business and LinkedIn in its financial reports.
Currently, the robotics and consumer hardware departments operate independently from other OpenAI businesses and report directly to Altman. These operations are highly confidential, with some internal staff describing them as standalone startups within the company.
OpenAI formally established its consumer hardware division in May of last year through a $6.5 billion stock acquisition of AI company io, founded by former Apple Chief Designer Jony Ive, and absorbed its team of approximately 55 people. Altman has internally shared details about an in-development hardware device: it is designed to be portable, pocket-sized, capable of sensing its environment, and is envisioned as becoming the third core smart device on a user's desk, alongside the MacBook Pro and iPhone.
OpenAI disclosed in legal filings this year that this hardware device will not be available to consumers before the end of February 2027.
Regarding robotics, OpenAI has been developing capabilities for several years, having previously trained a robotic hand to solve a Rubik's cube. Last year, the company entered a research partnership with robotics delivery firm CocoRobotics, in which Altman is also a personal investor.
Altman stated last month: "We are exploring how to go big in robotics. If there's one thing that could restore America's global competitiveness in manufacturing and physical industries, it would be building a massive number of robots and achieving scale in their self-research and production."
Comments