Korva is a scroll of the internet’s best ideas.
A year of research before investing
No template to follow
A software company comes with comparable businesses and a known recipe for what it takes to build one and what kind of founder succeeds, so a decision can come fast. A frontier company has no such comparisons.
Going native
The breadth of opportunity is wider, and the depth of understanding needed for each one is far greater, which makes the problem exponentially harder. It takes going native on the space: talking to researchers across the world, speaking with practitioners who would buy the technology, reading papers, and consulting people in the industry.
Why the effort pays
Being right pays off in a very big way, and there are many ways to be wrong, so the process is fundamentally more involved than for a company in the diffusion space.
Why do investors rely on comparable companies?
Comparable companies give investors a template for spotting winners, which matters because most venture investments lose money. Software startups in particular repeat patterns that investors have watched for decades. Correlation Ventures' analysis of more than 20,000 financings found that about 65 percent returned less than the amount invested.
Where do frontier breakthroughs first appear?
Many appear first as preprints on arXiv, a free repository run by Cornell University, often before formal review. The transformer architecture behind modern chatbots was posted there in June 2017 in a paper titled Attention Is All You Need. Eight Google researchers wrote it.
Why do frontier companies take so long to pay off?
Hardware and science problems must be solved before a product exists, and each failed attempt costs real money. SpaceX was founded in 2002, and its first three Falcon 1 launches, in 2006, 2007, and 2008, all failed to reach orbit. The fourth flight reached orbit in September 2008.