Global startup investment hit $510 billion in the first half of 2026, Crunchbase News reported July 2 — a record that already exceeds the $440 billion invested across all of 2025, and blows past the prior half-year record of $375 billion set in the second half of 2021 by roughly 36%. The half split into two record-setting quarters on their own: Q1 2026 brought in $305 billion, and Q2 added $205 billion, the second-largest quarter ever recorded.
The defining fact of the half is concentration, not breadth. AI-focused companies captured more than 70% of all global startup capital in Q2, up from roughly 50% a year earlier. OpenAI and Anthropic alone raised a combined $217 billion in H1 — 43% of every venture dollar deployed worldwide during the period — with Anthropic's roughly $65 billion in Q2 alone accounting for nearly a third of that quarter's global total. Sixteen companies closed rounds of $100 million or more in Q2, totaling $108.6 billion, or 53% of the quarter's entire funding volume.
The run-up traces back to the ChatGPT-driven investment supercycle that began in late 2022, but what's changed by mid-2026 is scale: what were once headline-grabbing nine-figure rounds are now routine, and the handful of frontier labs and infrastructure players commanding the largest checks have pulled so far ahead of everyone else that Crunchbase's own data shows the rest of the venture market — consumer, fintech, biotech, enterprise SaaS outside AI — effectively fighting over the remaining 30%.
“AI-focused companies captured more than 70% of all global startup capital in Q2, up from roughly 50% a year earlier.”
The competitive landscape inside that AI share is itself narrowing at the top: OpenAI and Anthropic dominate the frontier-model tier, while capital-intensive infrastructure players like Together AI (fresh off an $800 million raise at an $8.3 billion valuation this week), CoreWeave, Crusoe and Lambda compete for the compute layer underneath them. Geographically, the US still commands the majority of dollars but its share slipped from 83% of global capital in Q1 to two-thirds in Q2, with eight of the quarter's billion-dollar rounds based in the US against four each in Asia and Europe — a modest but real signal of capital beginning to diversify again.
The numbers read differently depending on which slice you examine. Late-stage funding rose 141% year-over-year in Q2 to $134 billion, evidence that investors are chasing proven AI winners rather than diversifying into new bets. Seed-stage capital, by contrast, barbelled: $2.8 billion went into mega-seed rounds above $100 million, while $5 billion spread across rounds of $10 million or less — a sign that even at the earliest stage, capital is polarizing between AI-anointed breakouts and everything else. Exit activity, historically the market's weak link since the 2021 peak, finally caught up: 32 companies IPO'd above $1 billion in Q2 and 24 were acquired above that threshold for a combined $113 billion, the largest M&A quarter on record, led by SpaceX's own $75 billion IPO (closing near a $1.77 trillion valuation) and its $60 billion all-stock acquisition of Cursor.
For founders, the practical read is stark: raising outside the AI frontier-lab and infrastructure tier is getting relatively harder even as the total pool of capital grows, because so much of the incremental dollar is flowing to a small number of already-massive rounds. For GPs, a market where two companies eat 43% of global funding is a market where fund-level returns increasingly depend on access to a handful of marquee names rather than broad-based sourcing. For LPs, the exit data is the genuinely encouraging half of the story — a record M&A and IPO quarter means the multi-year backlog of unrealized venture value finally started clearing in a real way.
The bear case is that this concentration is exactly the kind of setup that precedes a correction: if AI infrastructure spend or frontier-lab revenue growth disappoints even modestly, the capital concentrated in a handful of names has much further to fall than a diversified market would. It's also worth asking whether a $217 billion combined raise for two companies is capital markets pricing genuine capability gains or simply the last, biggest phase of an arms race that has to eventually slow.
What to watch: whether Q3 funding data shows the AI concentration easing or deepening further, whether the pickup in Asia and Europe's share of billion-dollar rounds continues, and whether the record exit pace (IPOs and M&A) holds up as more AI-native companies — Anthropic among them — move toward their own public listings later this year.