Two hundred and sixty-three mega-rounds took 81% of all venture funding in the second quarter of 2026. One company alone took nearly 40%.
That is the number that should have led the week. Instead the headline that traveled was a different one: 195 new unicorns minted in the first half of the year, already past the 193 the industry produced in all of 2025.
Put those two facts side by side and the celebration curdles.
Because in the same window that produced a record class of billion-dollar startups, CB Insights recorded deal count at its lowest point in over a decade, and new unicorn births falling more than two-thirds in a single quarter.
Two consecutive quarters above $200 billion — among the strongest stretches ever measured — sitting on top of the thinnest dealmaking in ten years. The trade press read that as breadth. It is the opposite. The record isn't width. It's altitude.
Two consecutive quarters above $200 billion sitting on top of the thinnest dealmaking in ten years. The record isn't width. It's altitude.
A unicorn is now something that can be conjured without a functioning market underneath it. The two metrics have decoupled. You can mint more billion-dollar valuations than ever while doing fewer deals than at any point since 2015, because a single frontier lab raising at $50 billion drags a whole cohort past the $1 billion line without touching the thousands of companies that used to define a rising market.
Look at how those unicorns were made. DeepSeek arrived at a $50 billion valuation on its first external financing. Etched doubled to $10 billion six months after its prior round. Hadrian went from $1.6 billion to $7.9 billion in seven months. Valar Atomics tripled to $6 billion in four.
These are not the tracks of a widening ecosystem. They are the same small pool of capital cycling back through the same names at ever-higher marks — nineteen of H1's new unicorns raised fast follow-ons, often inside six months, doubling their prior valuation.
Carta's data team, watching from inside the cap tables, put it plainly this month: capital is concentrating into fewer, larger deals across most stages. Round counts flat or falling. Capital raised rising sharply, especially late.
Crunchbase's own numbers show how far this has run. In 2025, 70% of U.S. funding — over $200 billion — went to 389 companies that raised $100 million or more. Ninety billion of that went to six companies.
The remaining 30% was scattered across roughly 6,000 companies. By early 2026, 80% of U.S. investment had gone to rounds of $500 million and up, across just 29 companies.
That is the ground truth beneath the unicorn count. Now the mechanism that will make it permanent.
For years venture has been waiting on liquidity. Paper marks don't return capital; only exits do. SpaceX's $85.7 billion IPO — the largest venture-backed exit of all time — cracked the window, and behind it sit OpenAI, Anthropic, Databricks and Stripe.
The reflexive assumption is that this wave rescues everyone: distributions flow back to LPs, LPs recommit, and the money washes down through the ecosystem to the next generation of funds and founders.
Andrew Gershfeld of Flint Capital argues the exact reverse, and the arithmetic is on his side.
When distributions land, investment committees rebalance and recommit — but they recommit first to managers with proven track records, and only later to emerging firms. The National Venture Capital Association's numbers show where that instinct leads: the 10 largest U.S. venture funds took nearly a third of all capital raised in 2025, while first-time fund formation fell to its lowest level in more than a decade.
Andreessen Horowitz alone raised over $15 billion across five funds last year — more than 18% of every venture dollar raised in the United States.
So when the AI IPOs print, the LPs receiving those distributions will hand a disproportionate share of the proceeds back to the firms that were already largest. Stronger distributions leave the biggest firms better positioned to raise still-bigger successors.
"This is not a liquidity flywheel," Gershfeld writes. "It is a concentration flywheel." Successful exits fund larger successor funds, which reinforce their own advantages, and around it goes. "The market may become larger without becoming broader."
That last line is the whole thesis. A $15 billion fund does not behave like a $500 million fund. It needs meaningful ownership and multibillion-dollar outcomes, so it leads larger rounds, pays higher prices, and defends its stake across financings.
Which is precisely why 263 rounds take 81% of the money. The fund structure and the deal structure are the same structure.
The tempting objection is that Anthropic's and OpenAI's scale expands the map for everyone. NEA's Madison Faulkner says she's "extremely excited about betting on seed and Series A." Dell's Daniel Docter points to "so much white space" around the giants.
Maybe. But that is a bet on a future that runs against the observed physics — and the physics are not a Sand Hill Road artifact. They are global.
Look at Africa. In H1, African startups raised roughly $1.4 billion, flat against the prior year. Flat dollars, but deal count fell sharply — meaning the same money went into fewer, bigger checks, all of them later-stage.
The reason is the same one driving the barbell in the U.S. "Exit discipline has become a credential rather than a footnote," says Ibrahim Sagna of Silverbacks Holdings. "Fewer, better vetted bets" define the cycle.
African fund managers, says 4Di Capital's Justin Stanford, face "a lot more scrutiny in terms of cash returns, not just paper performance" — pushing them later-stage, and pushing early-stage founders out of the frame. Both 4Di and Norrsken22 are now eyeing secondaries, buying existing stakes rather than funding new companies.
India shows the same tremor at a different frequency: startup funding there more than halved in July after a June spike. Volatile, concentrated, top-heavy.
From Lagos to Bangalore to Menlo Park, the same instruction: back the proven, price the winners up, starve the seed. Exit discipline as a credential is just the concentration flywheel spoken in a local accent.
Which leaves the uncomfortable question for LPs and emerging managers, the one the 195-unicorn headline was built to obscure.
Paper unicorns do not return capital. A record class of billion-dollar valuations, minted through fast follow-ons among a handful of names, is not a distribution — it's a promise contingent on public prints that haven't cleared yet.
And when those prints do clear — when Anthropic's secondary mark finally meets a public tape — the liquidity they release will not fund the next generation of funds. It will re-up the last one.
First-time fund formation is already at a decade low before any of this money has landed. The wave everyone is waiting for won't reverse that. It will deepen it.
H1 2026, then, is not the moment venture got wider. It's the moment it stopped trying. The market is getting taller, and taller is not the same as bigger, and neither is the same as open.