Du Jun, co-founder of Huobi and ABCDE, shared Sam Altman's old speech about the necessity for founders to continuously generate new ideas, stating that over the past decade, his greatest learning experience has been in trying to create entrepreneurs, all of whom ultimately failed.

Altman mentioned in a 2018 Y Combinator investor school that while a founder can be given an idea to start a company, the challenge is that the company must come up with new ideas almost weekly, making drastic adjustments and directional changes. Y Combinator conducted an experiment funding 20 strong teams that lacked ideas, and all failed.

Altman's conclusion is that good founders are those who continuously generate ideas, possessing intelligence, creativity, independent thinking, and the ability to view problems differently and conceive things that should exist but do not. Without this continuous flow of ideas, even the best execution teams cannot sustain the pressure of weekly iterations with a ready-made idea.

Du Jun related this to his own experiences over the past decade: it was not a lack of money, traffic, exchange access, or media exposure, but rather repeated attempts to "manufacture entrepreneurs"—providing direction, resources, and scenarios—resulting in no successful samples. He has simultaneously managed exchanges, media, funds, and compliance risk control platforms, covering a path from being a product manager at Discuz, to Tencent, co-founding Huobi in 2013, and later ABCDE, Vernal, and compliance infrastructure.

This type of failure structurally recurs: the "created" founders can launch the first version of a product but cannot generate the next week's crazy new ideas in response to market feedback, regulatory changes, liquidity exhaustion, or narrative shifts. The result is projects that can be narrated and financed but cannot be continuously reconstructed.

In market mechanisms, what is bought is "incubatable execution capability," and what is sold is "sustainable idea flow." Funds flow from exchange fees, fund LPs, and cash cow assets like domain names to designated entrepreneurial teams; benefiting those who already possess high idea flow, amplifying their iterations through existing channels; while the burden falls on founders treated as mass-producible objects and the investors who pay for their time and capital. This is an event-driven cognitive correction, not a new fundraising cycle.

For reference: Altman wrote about the same experiment in his Idea Generation article—almost all teams in the no-idea track failed; he later emphasized that once a project is elevated to a "company," ideas cannot be too crazy, easily turning into seemingly reasonable derivatives. Du Jun's ten-year sample and YC's 20-team sample point to the same filter.

Source: Public Information