Investors poured $297 billion into roughly 6,000 startups worldwide in the first quarter of 2026. That figure exceeds every full-year venture capital total before 2018. It equals nearly 70% of everything deployed across all of 2025. And four companies absorbed the majority of it.
OpenAI closed a $122 billion round. Anthropic raised $30 billion. Elon Musk’s xAI pulled in $20 billion. Waymo secured $16 billion. Those four companies alone collected $186 billion, or 64% of total global venture activity for the quarter, according to Crunchbase data published on April 1.
The numbers describe something more specific than a boom. They describe a structural rearrangement of how private capital flows.
The Math That Matters: One Quarter Versus All of History
AI startups captured $242 billion of the quarter’s total, representing 80% of all global venture funding. In Q1 2025, AI’s share stood at 55%. The jump from slightly over half to four-fifths in twelve months marks one of the fastest sector-concentration shifts in venture history.
The geographic concentration is equally extreme. U.S.-based companies raised $250 billion, or 83% of global venture capital. That is up from 71% in Q1 2025, which was already above historical averages from the decade before 2024. China followed at $16.1 billion. The United Kingdom came third at $7.4 billion. Both markets grew year-over-year, but the U.S. share grew faster.
Late-stage funding drove the distortion. A total of $246.6 billion flowed into 584 late-stage deals, up 205% year-over-year. Of that, $235 billion went to just 158 companies that raised rounds of $100 million or more. The top of the funnel is a pinhole. The bottom is a fire hose.
Why This Is Not a Normal Funding Cycle
Previous technology booms spread capital across hundreds of companies building similar products. The 1999 dot-com peak scattered money into thousands of web startups. The 2021 ZIRP boom funded an entire generation of fintech, crypto, and SaaS companies. Both cycles distributed risk across a wide portfolio.
Q1 2026 did the opposite. Four of the five largest venture rounds ever recorded closed in a single quarter. The concentration ratio (64% of all global VC to four companies) has no precedent in Crunchbase’s dataset. Even during the dot-com peak, no single quarter saw more than a third of total capital flow to fewer than five recipients.
The reason is structural: training frontier AI models requires physical infrastructure at a scale that software companies never needed. AMI Labs raised $1.03 billion just to build JEPA, a research-first company with zero revenue and zero product. OpenAI committed to spending over $100 billion on compute infrastructure in 2026. These are not software companies that can scale on AWS credits. They are building physical plants that cost billions before generating a dollar of revenue.
This makes the current cycle simultaneously a software story and an infrastructure story. Capital is flowing into data centers, custom silicon, power generation, and cooling systems alongside model development. TD Cowen estimates that Oracle alone plans to spend $156 billion on AI infrastructure, a sum large enough to force the company to lay off up to 30,000 employees to fund the buildout.
What the Seed and Early-Stage Numbers Actually Show
The megarounds dominate the headline, but the early-stage data tells a different story. Seed funding rose 30% year-over-year to $12 billion. That sounds healthy until you check the deal count: it fell 30% to 3,800 rounds. Fewer companies raised seed money, but the ones that did raised more of it.
Early-stage funding (Series A and B) totaled $41.3 billion across 1,800 deals, up 41% year-over-year. Series A grew. Series B declined quarter-over-quarter but stayed positive year-over-year. The pattern suggests that investors are concentrating bets earlier, writing bigger checks into fewer companies at the seed and Series A stage.
Median seed post-money valuations hit an all-time high of $24 million in late 2025, up from $18 million a year earlier, according to Carta data. AI startups command a 42% valuation premium over non-AI peers at the seed stage. The message from the data is blunt: if you are building something generic, the fundraising math does not work.
The IPO Market Did Not Follow the Money
Record private investment did not translate into a stronger IPO market. The U.S. market for new listings actually slowed in Q1 amid a broader stock market selloff in software. China’s IPO market picked up instead: 13 of the 21 venture-backed companies that exited above $1 billion were Chinese, including two AI foundation model companies (Z.ai and MiniMax) that debuted on the Hong Kong Stock Exchange at valuations above $6 billion each.
The largest IPO globally was Japan-based PayPay, a mobile payments fintech valued at $10 billion on listing. Not an AI company.
M&A provided a partial counterweight. Startup exits totaled $56.6 billion, making Q1 the third-strongest M&A quarter since the 2022 downturn. The largest deals were Savvy Games Group’s planned $6 billion acquisition of ByteDance’s gaming platform Moonton and Capital One’s planned $5.15 billion purchase of fintech startup Brex.
The disconnect between record private investment and a flat public market creates pressure. Companies holding unprecedented amounts of private capital need liquidity events. OpenAI is reportedly targeting an IPO by late 2026 at near $1 trillion. xAI-SpaceX is targeting June 2026 at up to $1.5 trillion. Anthropic, now approaching $19 billion in annualized revenue, faces growing IPO expectations. If those listings underperform, the $297 billion quarter becomes the ceiling, not the floor.
What the Concentration Ratio Breaks
When 64% of all venture capital flows to four companies, the remaining 5,996 startups split $111 billion. That is still a large number by historical standards. But it represents a fundamentally different market than the one that existed two years ago.
Talent flows toward the megaround recipients. The frontier model race between OpenAI, Anthropic, and Google DeepMind has already driven AI researcher salaries above $1 million annually at top labs. Startups outside the top four compete for the same researchers with a fraction of the capital.
Supply chains tighten. GPU allocation, data center leases, and power purchase agreements increasingly go to the highest bidder. Smaller AI companies building on the same infrastructure face longer wait times and higher costs.
Investor attention concentrates. Limited partners allocating to venture funds increasingly evaluate managers on their access to megaround deal flow, not their seed-stage portfolio construction. Fund formation data from Q1 2026 shows the top five VC closes accounted for $35 billion, more than half of all U.S. venture capital raised in the entirety of 2025.
The Fragility Question
Venture analysts who spoke to Crunchbase cautioned that the inflows increase systemic fragility. Companies carrying heavy fixed costs for specialized hardware and data centers cannot scale down quickly if funding slows. An abrupt regulatory shock, a shift in enterprise AI adoption rates, or a correction in the public tech market could disproportionately hit the companies that absorbed the most capital.
The proximate stress tests are already visible: chip supply chains and compute pricing remain constrained. Talent wage inflation shows no sign of slowing. And the profitability models underpinning compute-heavy businesses remain largely unproven at the scale these companies now operate.
OpenAI generates $2 billion monthly in revenue. Anthropic approaches $19 billion annualized. Those are real numbers. But the capital requirements to maintain frontier model competitiveness are growing at least as fast as revenue. The question is not whether these companies can generate revenue. It is whether revenue grows faster than the infrastructure costs required to stay competitive.
A single quarter worth $297 billion has made that question considerably more urgent.
This article is editorial analysis for builders and founders. It is not financial advice.