Global Startup Funding Hit a Record $510B in H1 2026, and Two AI Labs Took 43% of It

July 6, 2026. Global venture funding reached a record $510 billion in the first half of 2026, more than the $440 billion invested in all of 2025 and the largest half year total ever recorded, according to Crunchbase data published July 2. The defining fact inside the record: OpenAI and Anthropic alone accounted for $217 billion, 43% of all startup funding in the half. For anyone building on AI, buying AI, or raising for an AI product, this report is the clearest picture yet of the market you are actually operating in.

The key developments

1. H1 2026 totaled a record $510 billion, surpassing the previous half year peak of $375 billion from H2 2021. Q1's $305 billion was the largest quarter on record and Q2's $205 billion, into more than 5,000 startups, was the second largest.
2. OpenAI and Anthropic took $217 billion, 43% of the global total. Anthropic raised $65 billion in Q2 alone, close to a third of the quarter, and became the most valuable private company on Crunchbase's Unicorn Board, passing OpenAI after SpaceX exited via IPO.
3. More than 70% of Q2 startup capital went to AI focused companies, up from just under 50% a year earlier, and 16 companies raised billion dollar rounds in Q2 totaling $108.6 billion, 53% of the quarter.
4. Liquidity returned in force: Q2 set records for both venture backed IPOs and acquisitions. SpaceX went public at a $1.77 trillion value, raising $75 billion in the largest venture backed IPO ever, then agreed to acquire Cursor maker Anysphere for $60 billion, the largest startup acquisition ever. 32 companies went public above $1 billion and 24 were acquired at or above $1 billion, totaling a record $113 billion.
5. Concentration extends to geography: two thirds of Q2 capital went to US based companies, and late stage funding hit $134 billion in Q2, up 141% from a year earlier, while global seed totaled $12 billion.

What it means for operators

Three practical readings. If you buy AI rather than raise for it, the concentration is working in your favor: hundreds of billions in lab funding is subsidizing the tools you rent, which is why frontier capability keeps getting cheaper (Claude Sonnet 5's introductory pricing, half rate GPT-5.4 agent runs, and the price moves we covered in the Sonnet 5 deep dive). The rational play is to build on the subsidized platforms now while keeping the stack model agnostic. If you are a founder raising for a thin AI application, the honest news is that you are competing for the 57% of capital two labs did not take, and investors have watched infrastructure eat the value of undifferentiated wrappers. What still gets funded, and what the record M&A market now pays for, is owned distribution, proprietary workflow depth, and real revenue. That is a build discipline question before it is a fundraising question, and it is where a partner who has shipped products, from MVP to production, earns its keep; see our launch your SaaS service for exactly that path. And if you sell implementation rather than software, this is your market: record capital pouring into platforms creates a matching wave of businesses that need those platforms wired into their operations, which is the core work of an AI automation agency. The IPO cluster we have tracked since the Anthropic S-1 filing also matters here: a functioning exit market keeps the whole cycle liquid, which keeps the tool subsidies flowing.