OpenAI is quietly experimenting with a new kind of accelerator, though calling it an accelerator may be misleading. The Grove Program, launched in San Francisco, sits somewhere before the first spark of a startup, before a product roadmap, before even a pitch deck. Its focus is on people, not companies, and on ideas that haven’t yet crystallized. Grove is OpenAI’s attempt to catch raw talent at the pre-idea stage and immerse them in a five-week cohort designed to shape what those ideas could eventually become.
The design of the program is telling. Fifteen participants make up the first cohort. The journey begins in person at OpenAI’s headquarters, where the group spends its first week immersed in the culture, the tools, and the pace of frontier AI work. Then the rhythm shifts to a hybrid format — weeks of asynchronous collaboration, exploration, and mentoring, before reconvening in San Francisco for a final in-person close. It’s not an MBA crash course or a startup boot camp. Instead, it is more like a residency for people who think they might want to build with AI but are still searching for their problem set.
What OpenAI offers is less about cash and more about context. Participants are given mentorship from OpenAI’s own researchers and engineers, access to internal resources, and structured workshops that push them to probe questions rather than settle on answers too quickly. The point is not to churn out ten identical startups with generative AI veneers, but to expand the space of what might be possible by giving people the room to try, stumble, and iterate before committing to a path.
In many ways, Grove is a hedge against the brittleness of the current AI startup pipeline. Most accelerators want fully formed teams with business plans. Grove wants curiosity, technical hunger, and the willingness to experiment. That inversion shifts power back to individuals who might otherwise never make it past the gates of traditional venture filters. For OpenAI, it also creates a channel to engage directly with the next generation of AI builders, shaping the culture and direction of projects before they hit the market.
There are still unanswered questions. It isn’t clear what Grove participants are expected to produce by the end of the program, whether the arrangement comes with funding, or if OpenAI takes any stake in the intellectual property that emerges. Nor is it obvious how inclusive it can be beyond San Francisco, given that the opening and closing weeks are firmly anchored at OpenAI’s headquarters. Yet, the very act of launching a program like this reveals how OpenAI sees its role shifting: not just as a model-builder, but as an ecosystem cultivator.
For the broader AI world, Grove hints at a new type of institutional intervention. Instead of waiting for the fully armed startup to come knocking, OpenAI is attempting to lower the barrier to entry for people whose work may otherwise never leave the sketchpad. If accelerators have historically been about compressing time for companies, Grove is about stretching time for individuals — providing a pause, a buffer, a protected space where the next set of ideas might take root. Whether that produces breakthroughs or dead ends remains to be seen, but the gamble is clear. In an industry obsessed with scaling models, OpenAI is betting on scaling minds.
Read more here: https://openai.com/index/openai-grove/
The Grove Experiment: OpenAI’s Bet on Building Minds Before Companies
