Korva is a scroll of the internet’s best ideas.
Startup waves break the old playbook
The marketplace wave
First came Facebook, Twitter, and Dropbox. Peter Thiel even predicted that the next Mark Zuckerberg wouldn't build a social network. The next winners were Airbnb, Uber, and Lyft, which ran real-world marketplaces city by city. Those founders put out free donuts to attract drivers, set up incentives for them, and threw parties in every city to make the thing work.
The crypto wave
Crypto came next, and the people who got rich were the ones who understood what money even is, what fiat currency is, and what the double-spending problem is. Bitcoin wasn't even a company, so the question of how to invest in it had no familiar answer.
AI labs and defense
AI labs rewrote the pattern again. OpenAI started as a nonprofit research entity, Anthropic followed, and they became the big winners. Before that, hardly anyone in Silicon Valley was investing in or starting a research lab. Defense tech followed a similar path: showing up at a startup meetup in a suit and tie once drew jokes about being a cop, until Palmer Luckey made defense cool.
Why did ride-hailing have to launch city by city?
A marketplace works only when both sides are present in the same place, because a driver in one city is no use to a rider in another. Economists call this a local network effect. Each new city starts at zero, so the company has to recruit drivers before riders have a reason to open the app. That is why early ride-hailing companies spent heavily on incentives and local events in every market.
How did Bitcoin solve the double-spending problem?
Digital files can be copied, so a digital coin could be spent twice unless someone keeps the books. Satoshi Nakamoto's 2008 paper replaced the bank with a public ledger called the blockchain, which records every transaction in order. Computers called miners compete to add the next block by solving a difficult math puzzle, a method called proof of work. Rewriting a past transaction would mean redoing that work for every block after it.
Why were research labs rare as startups?
Research labs spend years and large sums before they sell anything, which does not fit the venture model of quick user growth. OpenAI was founded in December 2015 as a nonprofit research lab, so early donors were not promised a financial return. Training large AI models later required enormous amounts of computing power, and that cost pulled in investors at a scale few startups had seen. The pattern shows how a cost structure can decide who competes.
Why was defense work once avoided in Silicon Valley?
The region's early growth came from military contracts, including radar, microwave tubes, and satellites built by firms around Stanford and Sunnyvale. Later generations of tech workers saw consumer software as the future, and some pushed back against military work. In 2018, thousands of Google employees signed a letter protesting the company's role in Project Maven, a Pentagon program that used AI to analyze drone footage. Google then said it would not renew that contract.
More from My First Million
