Sources: last summer, Sam Altman reached out to at least one rocket maker, Stoke Space, for OpenAI to secure a controlling stake; the talks are no longer active
dude is like elon whose procrastination is seen as entrepreneurship [embedded post] Dare Obasanjo / @carnage4life : There was a time I wondered if OpenAI was trying to do too much at once and I don't wonder anymore. — They got overconfident and took their eye off of the ball. Wanting to compete with TikTok (Sora), Chrome (Atlas), the iPhone and SpaceX meanwhile Google caught up to their main product.
Context & Ripple Effects
The reported outreach places a possible rocket-maker acquisition alongside OpenAI’s broader push to secure computing capacity, including Altman’s dealmaking for vast compute. It suggests the company was at least evaluating infrastructure-adjacent expansion beyond its core model business.
Later reporting that OpenAI considered separating its robotics and consumer-hardware units points to an unresolved organizational question: whether adjacent bets should sit inside the lab or gain room outside it.
First-order effects
- Stoke Space does not gain OpenAI as a controlling owner because the reported discussions are no longer active.
- OpenAI avoids immediately taking on the governance, capital-allocation and execution demands of controlling a rocket manufacturer.
Second-order effects
- The halted talks leave OpenAI’s infrastructure strategy centered on partnerships and compute procurement rather than an immediate ownership move into space hardware.
- For Stoke Space and comparable infrastructure startups, the episode underscores that frontier-AI labs may be potential strategic counterparties, but exploratory interest is not equivalent to committed capital or control.
Third-order effects
- If frontier labs continue pursuing compute, hardware and other infrastructure adjacencies, their scope will increasingly test whether centralized lab structures can manage businesses with very different capital needs and operating cycles.
- The more durable pattern may be selective vertical integration: labs will weigh ownership against partnerships, particularly when expansion could distract from model development or complicate financing.
The trend: This is one data point in frontier AI labs probing control of strategic infrastructure while reassessing how far their corporate boundaries should extend.