He calculated exactly when video sites should profit—yet never built YouTube himself
In 2003, he calculated that video sites' bandwidth costs would only make business sense around 2005–2006—perfectly timed with YouTube's founding—but because he never actually built it himself, he handed off a company now worth $300 billion to someone else.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Betting on a decade-long trend, not a two-year cycle
Ryan says building a significant company requires betting on a ten-year trend—hot niches burn out in two or three years, and by the time you react, you've missed it. DoubleClick bet on internet advertising, MongoDB on unstructured data (a twenty-year trend), psychedelics are a five-year-old mental health wave whose full effects won't be clear for a decade, and nuclear power is a new sector he entered only two and a half years ago. The pattern: these bets were all contrarian at the moment he made them.
— Kevin RyanHe nailed the timing but never built YouTube himself
In 2003, his team modeled the bandwidth cost: content delivery cost $10 per thousand views while advertising brought in only $1—the unit economics didn't work. But they projected that by 2005, 2006, costs and ad revenue would converge. ‘We should have started building YouTube, but we didn't.’ YouTube was founded in 2005, and it's now worth $300 billion. He calls this 'second-order thinking'—not just asking what a trend will do, but what downstream effects that trend will trigger.
— Kevin RyanExiting before building a moat became essential
Gilt pioneered flash sales—year two revenues hit $175 million, year four reached $500 million. But designer brands launched their own e-commerce, department store websites followed suit, Farfetch entered the market, and the category commoditized. He offers an example: Theory had 20,000 units of end-of-season inventory to clear; Gilt could buy 1,000 of those units—no leverage against the brand. The board wanted to sell for $400 million. He judged it ‘a falling knife’ and sold for $250 million to Saks. Three years later, Saks tried to sell it back to him for $5 million.
— Kevin RyanLeaders your age may lose sight of trends as they unfold
In 1995, his Dilbert website took off, and he hard-coded ads to monetize it in a year. He pitched his parent company to fund an internet division with millions of dollars. The decision-maker was smart, from a traditional media background, but refused because ‘we need to wait for the next internet.’ Ryan realized the executive had simply ‘stopped feeling what was happening in real time.’ The executive was then the age Ryan is now—61, 62. That prompted him to leave and join DoubleClick, which had been founded six months earlier.
— Kevin RyanMaster the narrow market first, then expand
AlleyCorp's playbook: start narrow, then expand. Gilt initially held only one flash sale per week for women's clothing, gradually adding a second, third, fifth per week, and only after a year added men's, kids', travel, and home. Business Insider started with three journalists covering only New York tech; it later added Wall Street, defense, and other beats. At its peak it had 600 journalists, but even then, the defense vertical got exactly six dedicated reporters.
— Kevin RyanA nuclear company he backed at $20M now values at $6B
Valar Atomics, a small modular reactor company, was found by his deep-tech team, not through his own network. Three years ago at the time of investment, it was valued at $20 million; in the latest round led by Sequoia, it reached $6 billion. He credits timing: back then the US government wasn't yet behind nuclear power. Now he thinks the Trump administration got both psychedelics and nuclear right, ‘and moved both forward dramatically.’
— Kevin RyanOne psychedelic drug's patent locked down the PTSD market for two decades
They chose to develop Methylone, a rarer MDMA analog, rather than MDMA itself because in trials Methylone could be dosed once a week while MDMA couldn't—it more severely depletes serotonin and carries higher toxicity. The company holds a patent covering three indications: PTSD, depression, and anxiety. For twenty years no one else can use it for these three conditions. But pushing each indication through FDA costs $200 million to $250 million, so they can only develop one first. They chose PTSD partly because veterans' issues play as untouchable across the political spectrum, and because 60% of trial participants were women—sexual assault is the primary cause of PTSD.
— Kevin RyanRunning three and a half years with no revenue proved MongoDB's long-term value
He compares a database to a pacemaker: no one wants to be the first to use a new pacemaker—everyone wants to wait until ‘a thousand people have used it for two years.’ So MongoDB had to give itself away free in the early years, generating almost no revenue for three and a half years. Even by year eight, revenue was only $40 million. But usage kept climbing and the product kept improving. Today MongoDB is worth about $300 billion—a study in contrast to his decision to exit Gilt at the right moment. Both faced hard years, but one needed to be sold and the other needed to be endured.
— Kevin RyanIn their own words · checked verbatim
First of all, 10 years is important because to build an important company, unfortunately, you can't do it in two or three years. It does take 10 years.
Kevin Ryan4:06
The reason there was no video in 2003 is because it costs about $10 per thousand to serve content and you could get about a dollar in advertising. So that model doesn't work.
Kevin Ryan13:33
i said we have a falling knife here and so we went out and we only got 250 million dollars sacks came in and paid for it and three years later they said i'll sell back to you for five million so better late than never
Kevin Ryan20:40
the guy i was talking to was very nice smart traditional media guy and he said no because we're going to wait for the next internet
Kevin Ryan23:47
I recognize it because I start with a business crush, meaning there's an idea and I can't stop thinking about it.
Kevin Ryan39:19
It was at a $20 million valuation three years ago. They just raised, it was announced publicly, from Sequoia at $6 billion. And they have been crushing it.
Kevin Ryan42:21
you can take meth alone once a week in our trials. You actually can't take MDMA once a week. It's a little bit more toxic. It depletes your serotonin a little bit more.
Kevin Ryan55:51
a database is like a pacemaker. If I said to you, I've got this startup pacemaker idea. Do you want to try it? And you're like, no. No. Call me when a thousand people have used your pacemaker for two years.
Kevin Ryan1:13:24
Figures
| DoubleClick sale price | $1.1 billion | 2:01 |
| Gilt year-two revenue | $175 million | 18:39 |
| Gilt year-four revenue | $500 million | 19:39 |
| Gilt sold to Saks for | $250 million | 20:40 |
| YouTube current valuation | $300 billion | 14:33 |
| Valar Atomics valuation three years ago | $20 million | 42:21 |
| Valar Atomics latest round valuation | $6 billion | 42:21 |
| MongoDB current valuation | ~$300 billion | 1:13:24 |
| Cost to push one indication through FDA | $200–250 million | 53:46 |
| Otsuka acquisition of Transcend | $700 million upfront plus up to $525 million in milestones | 3:05 |
Glossary
- Second-order thinking
- Asking not just what will happen to a trend itself, but what downstream effects it will trigger.
- Public benefit corporation
- A company legally chartered to commit a share of profits to public benefit beyond shareholder returns.
- QSBS
- A US tax benefit allowing founders to exclude capital gains on early-stage company stock held for the required holding period.
- Methylone
- An MDMA analog that can be dosed weekly and is being developed as a prescription treatment for PTSD and other psychiatric conditions.
How to listen
Serial entrepreneurs, early-stage investors, and operators wondering when to exit and when to endure.
47:15–49:05: discussion of ping-pong and hiring a coach, tangential to startup investing principles.