Global startup funding hit a record $510 billion in the first half of 2026, according to Crunchbase data, comfortably surpassing the $440 billion invested across all of 2025. The headline is unambiguously bullish. The composition underneath it is not: OpenAI and Anthropic alone accounted for $217 billion of that total, or 43% of everything raised globally across every sector, every stage, and every geography combined.
That concentration shows up throughout the data. AI captured more than 70% of all global venture capital in Q2 2026, up from under 50% a year earlier -- a swing that would be remarkable even without the two-company skew. North American startups raised $392 billion in H1 alone, and nearly 90 new companies crossed the unicorn threshold this year, most of them AI-focused. Those numbers read as broad-based strength. They're only broad-based if you don't look at where the dollars actually landed.
“North American startups raised $392 billion in H1 alone, and nearly 90 new companies crossed the unicorn threshold this year, most of them AI-focused.”
The mechanism is straightforward: frontier labs need capital in quantities that dwarf normal venture rounds -- Anthropic's Series H-1 alone was $65 billion -- and every dollar committed to a $65 billion round is a dollar that isn't available for the next tier of AI infrastructure, application-layer, or adjacent-sector startups. LPs allocating to AI exposure through fund commitments are increasingly, indirectly, making a two-company bet even when their actual portfolio companies are elsewhere.
The practical effect for emerging managers and Series A/B investors: the "AI is eating venture" narrative is real at the top of the market and considerably less certain everywhere else. Deal volume for sub-$50 million AI rounds hasn't grown anywhere near as fast as dollar volume has, meaning more capital is chasing fewer, larger opportunities rather than a broadening set of investable companies.
What to watch: whether Anthropic's rumored October IPO and any subsequent OpenAI listing pull capital back out of private markets and toward public AI exposure, which would ease concentration at the venture stage even as it creates a new concentration problem in public indices.