AI startups raised $202.3 billion in 2025, according to Crunchbase’s year-end analysis. That figure represents a 75% increase from the $114 billion invested in 2024 and means AI captured nearly 50% of all global venture funding last year, up from 34% in 2024.
The concentration accelerated into 2026. In February alone, AI startups raised $171 billion — 90% of all global venture funding that month, per Crunchbase. Two companies accounted for most of it: OpenAI closed $110 billion (following a $40 billion round in 2025), and Anthropic raised $30 billion.
Mega-Rounds Dominate
CB Insights’ annual State of AI report put the full-year 2025 total at $225.8 billion, with mega-rounds of $100 million or more accounting for over 75% of all AI funding, as reported by Forbes. Foundation model companies alone raised $80 billion in 2025 — 40% of global AI funding — more than double the $31 billion they raised in 2024, per Crunchbase.
OpenAI and Anthropic together captured 14% of all global venture investment in 2025, according to Crunchbase data. The U.S. dominated, accounting for $159 billion — 79% of AI sector funding.
Capital Rotating to Vertical Applications
The more consequential shift is happening below the headline numbers. PitchBook’s Q4 2025 report found that niche, industry-specific AI applications overtook general platforms in both deal value and deal volume in the final quarter of the year, according to Forbes. CB Insights found healthcare led new AI unicorn creation in 2025, and Bessemer Venture Partners’ State of AI report argued that vertical AI has the potential to eclipse even the most successful legacy vertical SaaS markets.
For agent builders, the signal is clear: the foundation model race is consolidating around a handful of players with near-unlimited capital. The investable frontier is now vertical — agents built for specific industries with domain expertise baked in, not general-purpose platforms competing with OpenAI’s balance sheet.
What This Means for Builders
The funding data draws a line between two categories of AI company. On one side: infrastructure and foundation models requiring tens of billions in capital to compete. On the other: vertical applications that can leverage falling inference costs and open-source models to build domain-specific agents without raising a mega-round. The second category is where the deal volume is growing fastest. As Forbes noted, the well-funded infrastructure companies are spending millions to convince every industry that AI works — which means the market education is being subsidized for everyone building on top of them.