$1.3 billion. That’s how much money investors just handed to a company called Safe Superintelligence that has no product, no revenue, and exactly three employees.
Welcome to 2024’s venture capital space, where the largest seed rounds in history are going exclusively to AI companies. And I mean exclusively. Of the top 20 seed rounds closed in the past year, every single one went to a startup building something related to artificial intelligence.
The New Math of AI Investing
Traditional seed rounds used to hover around $2-5 million. Companies would use that money to build a prototype, find their first customers, and prove their concept worked. The whole point was to start small and grow carefully.
That playbook has been completely rewritten. AI companies are now raising $50 million, $100 million, even $200 million in their seed rounds. Poolside, an AI coding assistant, raised $126 million before launching publicly. Glean, which builds AI search for enterprises, pulled in $100 million at seed stage.
Why the astronomical numbers? Two words: compute costs.
Training AI models requires massive amounts of computing power, which means renting thousands of high-end GPUs from cloud providers. A single training run for a large language model can cost millions of dollars. Before you’ve written a line of customer-facing code, you might burn through $20 million just experimenting with different model architectures.
The Talent Arms Race
But hardware isn’t the only expense driving these mega-rounds. AI companies are competing for a tiny pool of researchers who actually know how to build these systems. We’re talking about maybe a few thousand people worldwide who have the right combination of skills.
These experts command salaries that would make a Wall Street banker blush. Total compensation packages of $1-2 million per year are common. Some top researchers are getting offers north of $5 million annually. When you need to hire 20-30 of these people just to have a functional team, you’re looking at a $50 million annual payroll before you’ve made a single sale.
Why Investors Are Comfortable With This
You might be wondering: isn’t this completely insane? How can investors justify giving hundreds of millions to companies with no proven business model?
Here’s their thinking. The AI market is growing so fast that being six months ahead of competitors could mean capturing billions in revenue. Investors believe we’re in a “winner-take-most” moment where the companies that move fastest will dominate their categories for years.
There’s also a fear of missing out that’s hard to overstate. Venture capitalists watched NVIDIA’s stock price increase 10x in two years. They saw Microsoft’s market cap jump by over a trillion dollars largely on AI enthusiasm. Nobody wants to be the firm that passed on the next OpenAI.
What This Means for Everyone Else
If you’re starting a company that isn’t AI-related, this trend creates real challenges. Venture capital is a finite resource, and when billions flow toward AI startups, there’s less available for everything else. Seed rounds for non-AI companies have actually gotten smaller over the past year, even as AI rounds have exploded.
The talent competition is equally brutal. That experienced engineer you want to hire? An AI startup just offered them three times your budget plus equity that might actually be worth something.
The Risks Nobody Wants to Talk About
Here’s what keeps me up at night: most of these companies will fail. Not because they’re bad or their teams aren’t talented, but because the math is unforgiving. When you raise $200 million at seed stage, you need to eventually build a company worth billions to generate returns for investors. That’s an incredibly high bar.
We’re also seeing companies raise huge rounds before they’ve figured out product-market fit. Having unlimited capital can actually be dangerous because it lets you avoid hard questions about whether anyone actually wants what you’re building. You can keep hiring, keep training bigger models, keep pivoting, all while burning $10 million a month.
The AI boom is real, and some of these companies will become enormously valuable. But when the dust settles, we’ll probably look back at 2024’s seed round sizes the same way we now view 1999’s dot-com valuations: as a moment when excitement outpaced reality, and a lot of smart people convinced themselves that this time was different.
đź•’ Published: