Why a Former OpenAI Employee Says Cash Out Before an IPO

Lead

Andrew Ho, a former member of OpenAI’s technical team, left the company and quickly advised former colleagues to take money off the table if they have access to a tender offer. His message was not simply about personal finance, although he reportedly still holds about $700,000 in OpenAI equity that he cannot immediately cash out. It was a broader warning about how differently private and public markets may value frontier AI companies.

Key points

• Private markets and public markets reward different stories. Ho’s central point is that private investors may be willing to pay for a distant AGI narrative, tolerate large losses and accept long time horizons. Once a company goes public, however, investors tend to scrutinize revenue growth, margins, cash flow and the path to profitability more directly.

• An IPO does not mean instant liquidity for employees. Even if a company successfully lists, employee shares are typically subject to lock-up periods. When those restrictions expire, concentrated selling by employees can create additional market pressure. In that context, a tender offer can be valuable because it converts paper wealth into actual cash sooner.

• Trillion-dollar valuations require enormous revenue. The article notes Ho’s rough reasoning: even under favorable assumptions such as high gross margins and a generous earnings multiple, supporting a trillion-dollar valuation would require very large annual revenue. Frontier AI labs, however, are not merely software firms serving inference at high margins.

• The cost race does not stop. To remain ahead, leading labs must keep funding next-generation training runs. If they slow down, rivals such as other closed labs, as well as lower-cost open-source models, can close the gap. This creates a “Red Queen” dynamic: companies spend heavily not necessarily to break away, but to avoid falling behind.

• Skepticism about valuation is not skepticism about AI. Ho’s view, as described in the source, does not reject the long-term importance of AI. Instead, it questions whether current valuations already assume a smooth path toward major capability breakthroughs. He also argues that high-quality training data may become a more decisive bottleneck as returns from scaling become less automatic.

Why it matters

The debate shows that the AI market is moving from narrative enthusiasm toward harder questions about monetization. Over the past few years, leading model companies have benefited from technological momentum, rapid product adoption and the promise of AGI. But the business model remains expensive: training costs are real and upfront, while the revenue created by each new model generation is less certain.

For employees, the lesson is that equity value on paper is not the same as liquidity. For investors, the question is no longer only whether AI will transform the economy. It is which companies can convert technical leadership into durable profit while competing against both well-funded rivals and cheaper open models.

The larger implication is that the AI industry may be entering a more selective phase. Frontier labs will continue to spend heavily on compute, data and talent. At the same time, companies building datasets, reinforcement learning infrastructure, evaluations, deployment tools and vertical applications may benefit from the next stage of the market. The OpenAI liquidity debate is ultimately a stress test for the commercial assumptions behind the frontier AI boom.

Source: QbitAI