The venture capital market just bifurcated into two tiers: AI startups and everything else. According to Carta data, AI startups captured $52.5 billion—41 percent—of the $128 billion in venture funding deployed in 2025. xAI raised $20 billion in January. OpenAI closed a $110 billion round in February. Anthropic hit $30 billion at a $380 billion valuation. These three companies alone account for more capital than the entire venture ecosystem deployed to healthtech, edtech, and climate tech combined.
The market just admitted what it believes: anything that isn’t AI is a second-order bet.
The Herd Has Moved—Everyone Else Is Running After
In 2020, the venture capital narrative was AI-assisted X. AI-powered diagnostics. Machine learning for supply chain optimization. These were valuable, differentiated stories. VCs funded them because the AI layer made a business 2-3x better.
That story collapsed when large language models got good enough to do interesting work out of the box. Suddenly, you don’t need AI-powered email. You need AI that can write emails. You don’t need machine learning for contract analysis. You need Claude to read contracts. The value moved from “how can we apply AI to our business?” to “how can we build AI that people want to use directly?”
Venture capital follows gravity. When everyone in the LP network started asking about AI exposure in their portfolio, every partner at every firm started tilting capital toward AI bets. The founders pitching non-AI businesses started hearing the same question: “Where’s the AI angle?” If the answer was “our product is better,” the meeting ended. If the answer was “we’re using Claude to do X,” the pitch continued.
This isn’t rational resource allocation. It’s herd behavior dressed up in data. And it’s already breaking the venture ecosystem.
What Happens to Healthtech When Venture Dries Up
In 2024, healthtech companies raised $8.3 billion across 217 deals. In the first quarter of 2025, that’s already down 30 percent. Not because healthtech got worse. Because every dollar a venture partner hasn’t deployed to an AI company yet feels like a mistake waiting to happen.
Same pattern across edtech ($4.2B in 2024, down 45% YTD), climate tech ($3.8B in 2024, down 35% YTD), and fintech ($12.1B in 2024, down 28% YTD). These are real industries with real problems. They’re solving harder technical problems than most AI startups. But they’re not AI, so they’re fighting for the 59 percent of capital that VCs are trying to deploy while avoiding the worst feeling in 2026: missing the next $100B company because you were writing checks to a climate startup instead.
The cruel part: some of those climate startups and healthtech companies would have been $10 billion outcomes if they’d gotten the capital. They won’t be because venture capital is too busy chasing the 0.001% chance of another xAI.
This Is a Bubble, But It’s Not the Kind That Pops Quickly
People keep asking: Is this sustainable? Will venture capital diversify back out? The answer is: for another 18 months, yes, this is sustainable. The AI upside case is so large that any rational actor has to chase it. If OpenAI becomes a $500 billion company, the three firms that funded it return enough to forgive a hundred bad healthtech bets.
The mathematics work because the upside is genuinely big. AI models might be the most important infrastructure investment since electricity. Some venture partner will get genuinely rich funding the right AI companies. That keeps the herd running.
But bubbles don’t pop because someone predicted them. They pop because of two things: (1) competition erodes returns, and (2) better opportunities appear elsewhere. We’re already seeing both. Every major cloud provider is building language models. The moat that made xAI and Anthropic valuable is eroding. And the capital required to stay competitive is so high that only a handful of companies will survive the next two years.
When the dust settles, venture will have blown $150-200 billion funding AI companies. Three to five of them will be worth something. The rest will be acquihires or forgotten. And every founder who wasn’t building AI in 2024 will have spent the last three years fundraising against the current.
What This Means for Non-AI Founders
If you’re building a real business outside of AI, here’s the hard truth: you are not getting venture capital on the original timeline. You are not getting the check size you would have received in 2023. You are not competing for partner attention fairly because your partner is already committed to two AI deals that need money.
You have three options: (1) bootstrap and grow slower, (2) find a strategic investor who cares about your space (corporate VC funds, strategics), or (3) find a thesis investor who went contrarian and is specifically trying to bet against the herd. Those investors exist. They’re usually older, they’ve seen manias before, and they’re patient enough to wait five years for a 100x return outside of AI.
The venture market isn’t broken. It’s just revealed something uncomfortable: when everyone is running the same direction, the exits are crowded and the roads are empty everywhere else.
