Two Companies Took 43% of Every Venture Dollar on Earth. Here Is What That Did to Your Seed Round.
In the first half of 2026, OpenAI and Anthropic raised $217 billion between them. That is roughly 43% of all venture capital deployed worldwide, into two companies.
Global venture funding hit $392 billion and set records. February alone saw $189 billion in startup funding, the largest single month ever recorded. Every headline about the venture market this year has been a story about abundance.
Now the number that describes the market most founders are actually in: seed deal count fell about 31% year over year.
Both things are true
This is the part that confuses people reading the trade press, so it is worth being precise.
Seed dollars rose roughly 30% year on year. Seed deals fell roughly 31%. Those are not contradictory figures — they are the same fact described from two directions. The aggregate went up because the average check got enormous, not because more founders got funded.
A single AI lab three months old raised a $480 million seed round at a $4.48 billion valuation. One round like that lands in the seed bucket and lifts the quarterly total past a hundred ordinary seed rounds that never happened.
When dollars rise and deal count falls, you are not looking at a recovery. You are looking at concentration. The aggregate is being set by a handful of outcomes at the top of the distribution, and the median founder's experience is moving in the opposite direction from the headline.
The check you actually need got harder to find
Here is the mechanism that matters, and it is not about AI at all.
Emerging managers — the smaller, newer funds that write the $500,000 to $5 million checks into the next generation of companies — saw their own fundraising fall 35% year over year, to roughly $12 billion. These are the funds that take a first meeting with a founder who does not have a warm introduction from a tier-one partner. They are the entry point to the entire venture ecosystem.
When LPs concentrate into established mega-funds chasing foundation-model allocations, emerging managers raise less. When emerging managers raise less, they write fewer first checks. When fewer first checks get written, the seed deal count falls 31% while the dollar total goes up, because the dollars went somewhere else entirely.
The funding market did not get worse. It got K-shaped. If you are building a frontier lab, capital has never been more available in the history of the asset class. If you are building a very good vertical software company with $40,000 in monthly revenue and a credible path, the market for your round has materially thinned — at the exact moment the press is telling you venture is booming.
That gap between the narrative and the lived experience is doing real psychological damage to founders right now, and it is worth naming plainly: the problem is probably not you.
What concentration does to the companies that do get funded
Two second-order effects we are watching, because they change how you should plan.
Valuations detached from traction. Seed-stage AI companies raised at valuations roughly 42% above non-AI peers this year. For a founder, a high seed valuation is not a prize — it is a hurdle you have to clear at Series A, usually within eighteen months, in a market whose appetite may have moved on. Companies that raised hot seed rounds in 2025 are walking into that conversation now, and the flat round is back.
The missing middle. Capital is available at the very top and, through accelerators and angels, at the very bottom. The gap has opened in the middle — the $1M to $5M institutional round for a company with real customers and no foundation model. That is where a large share of durable software businesses have always been born.
Why we build the way we do
We should be direct about our bias here: AW3 is a venture studio, so our answer to a thin seed market is structurally self-serving. Take the argument on its merits or leave it.
The studio model is not better than venture capital. It is a different response to the same question of how a company gets from an idea to a going concern, and it happens to be less sensitive to the thing that broke this year.
A studio supplies the team, the infrastructure, and the early capital from inside the operating company rather than sourcing each separately from a market that has just concentrated 43% of its dollars into two names. When the first-check market thins, a company being built inside a studio does not notice, because its first check was never going to come from an emerging manager who could not close their fund.
That is the entire structural claim. It is narrower than most studio marketing suggests, and we think the narrower version is the true one. We wrote the longer argument in The Venture Studio Model: A Guide for Founders, including where the model is a worse fit than raising a conventional seed — which it often is.
If you are raising into this
Practical read, from watching rounds happen around us this year:
- Stop benchmarking against the headlines. The $392 billion number has nothing to do with your round. Benchmark against companies of your stage and sector that closed in the last two quarters, and find those through founders rather than press releases.
- Assume the round takes twice as long. Deal count falling 31% means more meetings per close. Plan runway on the pessimistic case; a bridge negotiated from strength is a different conversation than one negotiated at four months.
- Court emerging managers early anyway. They write the first checks and they are under pressure, which means they are also hunting for the differentiated non-consensus bet that justifies their next fund. That can work in your favor.
- Do not buy the highest valuation offered. In a concentrated market the premium you take at seed becomes the hurdle you cannot clear at A. Price for the round after this one.
- Get to revenue earlier than the playbook says. Capital concentration makes default-alive a strategy rather than a virtue.
The abundance is real. It is just not evenly distributed, and almost none of it is pointed at you.
