Frontier artificial intelligence labs, like OpenAI, face unique challenges. While their founders believe their work has the potential to end the world, they also need to demonstrate immediate commercial success due to the high costs of developing capable AI models. OpenAI attempted to reconcile these tensions by combining a not-for-profit oversight board with a for-profit commercial entity that limited returns. However, it is now expected that OpenAI will abandon its non-profit status as investors in its latest funding round can recover their money if this conversion does not occur.
Last year, tensions between OpenAI’s non-profit board and co-founder/CEO Sam Altman resulted in Altman’s firing and subsequent rehiring. The board has since been replaced and several senior executives have left the company. With no profit cap in place, Altman now has more freedom to structure the company as he sees fit and raise necessary funds.
OpenAI was initially founded as a non-profit research organization in 2015 but created a for-profit subsidiary in 2019. Microsoft president Brad Smith praised OpenAI’s structure at the Paris Peace Forum last November, comparing it favorably to Meta where founder Mark Zuckerberg maintains control through shares with extra voting rights.
While investors like Microsoft may have accepted this unusual structure, it was not much of a sacrifice considering that OpenAI is not yet profitable and returns remain theoretical. Additionally, Microsoft would receive 75% of eventual profits until recouping its initial investment before any profit cap came into effect.
Moving forward, if OpenAI adopts the public benefit corporation model used by AI peers such as Anthropic and xAI, investors need not worry about lofty goals hindering profitability. This model requires companies to pursue social goals but does not impose strict compliance standards.
As generative AI products become increasingly lucrative, more teams are faced with deciding whether they prioritize philosophical debates or commercialization efforts. However, there is no inherent conflict between running a straightforward for-profit company and building safe technology; reckless behavior that stifles an industry jeopardizes profits.
Companies like Google DeepMind and Anthropic have more conventional structures than OpenAI but still prioritize AI safety through substantial dedicated teams. In this industry, complex corporate engineering achieves little progress.