Global AI venture funding hit a record $407 billion in the first half of 2026, and a striking share of that capital flowed to just two companies. OpenAI raised $122 billion in a single March round, while Anthropic closed $30 billion in February and $65 billion in May. Together, OpenAI and Anthropic captured more than half of all AI venture dollars deployed in the period, according to PitchBook data published by Yahoo Finance on August 11, 2026, underscoring how concentrated private-market conviction has become around frontier model labs.
A record pool, a narrower funnel
The headline number is the largest six-month venture total on record for any technology category. The structure behind it is more revealing than the size. PitchBook counted roughly 3,500 AI deals in H1 2026, down from 8,290 across the full year 2025, yet aggregate dollars climbed to new highs. Dimitri Zabelin, senior analyst at PitchBook, framed the pattern as concentration rather than cooling. In his view, the data shows investors are not retreating from AI; they are pulling back into a smaller set of names that can plausibly absorb nine-figure rounds and still move the industry forward.
Three rounds alone, the OpenAI March financing and the two Anthropic rounds, delivered an estimated $200 billion or more. That is a meaningful share of the $407 billion pool coming from a tiny fraction of the 3,500 transactions tracked in the period. The asymmetry between dollars and deal count is now the defining feature of the AI venture cycle.
Horizontal platforms versus vertical applications
PitchBook’s taxonomy sorts AI startups into two buckets. Horizontal platform companies, defined as the frontier labs plus adjacent infrastructure players such as World Labs and ElevenLabs, accounted for 70.8 percent of deal value, or roughly $288.1 billion. Vertical application companies, the firms building products and services on top of frontier models, made up 62.9 percent of deal count but only 12.9 percent of capital deployed.
The split captures a market that is willing to underwrite enormous platform bets while leaving most downstream builders to compete for smaller checks. Zabelin summarized the dynamic in a statement accompanying the report. The concentration of capital into frontier models, he said, shows the conviction investors have in OpenAI’s and Anthropic’s ability to not just scale their own platforms but the vertical applications built on top of them. He added that, despite the macroeconomic headwinds, it is still an extraordinary amount of capital.
The macro backdrop investors are underwriting against
That backdrop is unusually hostile for a record funding period. PitchBook cited Iran war-driven inflation, a pullback in expected Federal Reserve rate cuts, and tighter credit conditions as the principal headwinds shaping 2026 deal terms. Larger rounds at higher implied valuations have continued, but they have done so on shorter diligence cycles and with more structured protections than the 2024 to 2025 vintage of AI financings.
The implication is that the concentration is not just preference, it is also defensive. Investors appear to be choosing a handful of platforms they believe can survive a tighter funding environment, while letting the long tail of vertical application startups fight for follow-on capital on less favorable terms.
IPO calculus shifts as OpenAI weighs a 2027 listing
The concentration is also reshaping exit planning. OpenAI is reportedly considering delaying its initial public offering into 2027, a slippage attributed to internal benchmark misses on frontier capability targets and to global market volatility. A 2027 listing would be the first major public offering from a frontier AI lab and would set a reference price for every private holding in the category.
Anthropic has not announced an IPO timeline, though its February and May 2026 rounds were among the largest private financings on record. The two companies together now anchor the public-market narrative investors will use to price the next generation of AI listings. A delayed OpenAI debut would push that reference point further out and keep private marks elevated for longer.
Outlook: concentration as the strategy
What the deal count decline signals about conviction
The drop from 8,290 AI deals in 2025 to roughly 3,500 in H1 2026 is often read as a cooling market, but the dollar totals suggest the opposite reading. With $407 billion deployed across fewer transactions and more than half concentrated in OpenAI and Anthropic, investors appear to be substituting breadth with intensity rather than retreating from the category. The shrinking deal count, paired with rising aggregate value, implies that capital allocators are willing to sit out marginal opportunities to preserve room for the platforms they view as defensible. For limited partners, founders, and corporate strategics watching the second half of 2026, this combination of fewer deals and larger checks sets the tone for how remaining capital is likely to be rationed.
The H1 2026 data describes an AI venture market that is simultaneously the largest and the narrowest in the sector’s history. Capital is flowing at record pace into OpenAI, Anthropic, and a short list of horizontal platform peers, while the broader application layer is being funded with a smaller share of dollars across a larger share of deals. For limited partners and corporate strategics, the practical question is whether the platforms can keep absorbing checks at this scale without compressing returns for everyone else. For founders building vertical AI products, the data suggests the funding environment remains open, but only at rounds and valuations calibrated to a market that has chosen its winners early. As OpenAI and the wider frontier cohort head toward a possible 2027 public debut, the concentration visible in the H1 2026 numbers is likely to define the AI venture cycle for the remainder of the decade.