James Dyson built 5,127 prototypes of his bagless vacuum over five years before landing on the version that became the DC01. That is a long time to keep going on a product that did not exist yet.
J.K. Rowling spent about five years writing the first Harry Potter book. After her marriage ended, she moved to Edinburgh with her baby daughter and finished the manuscript as a single mother on benefits, writing in cafés while her daughter slept beside her in a pram. It took her agent a year to find a publisher, and Bloomsbury published Harry Potter and the Philosopher’s Stone in 1997.
Airbnb’s founders hit the same wall in 2008. Brian Chesky and Joe Gebbia were trying to raise $150,000 for 10 percent of the company and pitched seven Silicon Valley investors. Five said no. Two never answered. With money running out, they designed limited-edition cereal boxes, Obama O’s and Cap’n McCain’s, and sold them for $40 a box, raising around $30,000. Airbnb has confirmed the campaign kept the company running that year, and Y Combinator co-founder Jessica Livingston has said the founders were doing whatever they could because they were broke. Y Combinator itself invested $20,000 for 6 percent of the company that January.
None of these stories turns on talent or timing. None of the work paid for itself while it was happening. Dyson needed years of failed prototypes before one worked. Rowling needed a year of rejection before an agent found a publisher. Airbnb needed a cereal business to survive long enough to reach Y Combinator. The expensive part was staying alive, not making the thing.
The same pattern shows up at scale, even where exact figures shift. The Bureau of Labor Statistics tracks how long new business establishments survive, and that survival rate varies substantially by year, industry and region, with recession-era businesses faring particularly badly in some places. Kickstarter shows a similar spread: success depends heavily on category, and the platform’s own data shows first-time creators succeeding at markedly different rates than repeat creators.
Cheap tools have made the first version easier to produce. Recording a track, listing a product or launching a campaign now takes a fraction of the money and equipment it once required. What has not gotten cheaper is what comes after: testing it, improving it, finding an audience, and finding out whether anyone wants it. That stretch runs for months or years, and most projects stop somewhere inside it, not because the idea was wrong but because the money, time or energy to continue ran out.
AI tools are already speeding up parts of that stretch, from drafting code to producing early prototypes. They have not touched the other cost: staying alive long enough to make the next version. Someone still has to cover rent and groceries while the work earns little or nothing, whether that comes from savings, a day job, a partner’s income, public benefits, or a business selling novelty cereal. The finished product gets remembered. The runway is what got it there.
