Global Startup Funding Hits Record $510 Billion in H1 2026, Powered Almost Entirely by AI

The numbers coming out of Crunchbase’s H1 2026 venture capital analysis aren’t just records — they’re a category shift. Global startup investment reached $510 billion in the first six months of 2026, surpassing the entire $440 billion invested throughout all of 2025 in a single half-year window. Q1 alone generated $305 billion — the largest single quarter in venture history. Q2 added $205 billion, good for second place on that same all-time list.

The driver is singular and overwhelming: artificial intelligence. More than 70% of global startup capital in Q2 2026 targeted AI companies. And within AI, the money is even more concentrated than that headline suggests.

The OpenAI-Anthropic Gravitational Pull

OpenAI and Anthropic together claimed 43% of all H1 2026 venture funding globally. That figure deserves to be read twice. Two companies — competing in the same sector, at adjacent capability tiers — captured nearly half of all the startup investment on earth for six months.

The defining deal was Anthropic’s $65 billion raise at a $965 billion valuation, which single-handedly accounted for roughly half of Q2’s AI investment total. This is the largest private funding round in history, and it reflects investor conviction that the frontier AI market will be winner-concentrated enough to justify pre-IPO valuations approaching $1 trillion. The round gives Anthropic capital to compete on training compute, inference infrastructure, and enterprise sales at a level that few companies outside the Big Tech universe have ever attempted.

OpenAI, which raised a $40 billion round at a $300 billion valuation in early Q1 2026, also contributed to the headline distortion. Between the two companies, the concentration of AI capital is extraordinary — and it’s creating a bifurcated venture market where frontier AI labs receive once-unimaginable check sizes while the rest of the startup ecosystem competes for a substantially smaller pool.

North America Dominates, But the Gap Is Narrowing

North American companies captured $392 billion in H1 2026 — the vast majority of the global total. Q1’s $252.6 billion was the largest quarterly total of all time and more than triple the prior quarter’s figure. The United States captured roughly 83% of global Q1 funding before settling back toward two-thirds in Q2, as Asian and European investment activity recovered.

The regional distribution tells an important story about the AI investment geography. The U.S. dominance in frontier AI capital is not evenly spread across the startup economy. It is concentrated in a small number of very large companies — primarily the frontier AI labs and a handful of AI infrastructure providers — while early-stage deal volume remains well below the peak levels of 2021-2022.

In North America specifically, the Crunchbase analysis notes that deal counts stayed “significantly below historical highs despite unprecedented funding totals” — meaning more money went to fewer companies at larger check sizes. This is the defining characteristic of the current funding environment: a narrow pipeline of massive AI bets alongside a more constrained environment for earlier-stage companies.

The Exit Market Opens

H1 2026 also delivered the most significant exit activity in years. Q2 alone saw 32 companies complete IPOs at billion-dollar-plus valuations and 24 companies acquired at $1 billion or more, generating $113 billion in M&A transaction value.

The landmark exits include:

• SpaceX IPO: Completed in June 2026 at a $1.77 trillion valuation, raising $75 billion — the largest IPO in history by a significant margin. The SpaceX listing also created the liquidity event that enabled SpaceX’s subsequent $60 billion acquisition of Anysphere (Cursor).
• Cursor acquisition by SpaceX: At $60 billion, this is the largest startup M&A deal ever recorded. It happened within weeks of the SpaceX IPO, demonstrating how the most well-funded acquirers can move extremely quickly when strategic opportunities arise.
• Cerebras IPO: The AI chip design company completed its offering at a $5.6 billion valuation, providing a successful public market exit for investors who had been patient through multiple delays in the company’s IPO timeline.

The return of a functioning exit market matters for the venture ecosystem as a whole. For most of 2023-2025, the IPO window was effectively closed for growth-stage tech companies, trapping capital in portfolios and limiting the recycling of investor returns into new funds. The opening of that window in 2026 is a structural catalyst for the next cycle of early-stage investment.

The Stage Distribution Problem

Behind the record headline numbers lies a more complicated picture for early-stage founders. Late-stage venture dominated Q2 with approximately $134 billion — up 141% year-over-year. Early-stage (Series A/B) grew 100% year-over-year to reach $589 billion across the quarter. But seed-stage investment declined from Q1, falling to $4.9 billion.

This pattern reflects a fundamental tension in the current venture market. The largest LPs and most active funds are concentrating capital at late stage because that’s where the AI winners are crystallizing. Frontier AI companies reaching late-stage rounds are raising at valuations that were previously reserved for public companies, and the concentration of institutional capital chasing a small number of likely winners has pushed check sizes to historic highs.

The implication for founders earlier in the journey is asymmetric access. Getting from zero to seed has not fundamentally changed. But the path from seed to Series A — where institutional venture capital typically becomes the primary funder — is increasingly contested, because the funds that would previously have led those rounds are now deploying a larger fraction of their capital into established late-stage AI companies.

Non-AI Sectors: Defense and Healthcare Hold

Not all of the H1 2026 momentum was AI-driven. Defense technology and healthcare saw significant rounds as well. Anduril Industries raised a $5 billion Series H — reflecting the ongoing surge in defense technology investment that has accompanied both geopolitical uncertainty and the increased autonomy of AI-enabled weapons and surveillance systems. Prometheus, a physical AI startup applying machine learning to industrial automation, raised $12 billion — the largest physical-world AI round in 2026.

These rounds suggest that while frontier AI labs capture the headlines and the bulk of the capital, the AI investment boom has a second layer: companies applying AI to high-value physical domains where automation can unlock meaningful efficiency gains or capability enhancements in sectors like defense, logistics, manufacturing, and drug discovery.

What Comes Next

The H1 2026 record raises an obvious question: is this pace sustainable?

The bullish case is that the numbers make sense given the scale of the technology opportunity. The frontier AI market is competing for a share of global GDP at a moment when the technology’s applicability to knowledge work is accelerating. Trillion-dollar valuations for companies that might capture a fraction of that TAM aren’t obviously wrong even if they are uncomfortable.

The bearish case is that $510 billion in six months has created concentrated positions in a small number of companies whose ability to generate revenue proportional to their valuations remains unproven at the required scale. If any of the mega-rounds — Anthropic at $965 billion, OpenAI at $300 billion — fail to execute on the growth assumptions priced into their valuations, the unwinding could be severe.

What seems certain is that the structure of the venture market has permanently changed. Two AI labs account for 43% of global startup funding. The largest IPO in history was a space exploration company that immediately acquired the largest startup M&A target in history. Deal volume is declining while deal size is soaring. The rules that governed venture capital for the past two decades are being rewritten, and the rewrite is happening faster than most participants can track.