On October 18 PitchBook reported what a record year of first checks is doing to investors. It estimates 8,625 first-time VC financings in the US through the third quarter, above the 2021 record, while the number of active VCs has been declining for years. The result, in PitchBook's telling, is an "almost comically competitive" market in which smaller funds now court founders who haven't founded anything yet. The median pre-seed pre-money valuation has tripled to $12 million from $4 million in 2020, and the typical seed deal is priced at $19.5 million before the money.
The labels pre-seed and seed now cover Enact Intelligence's $30 million pre-seed, Dream Labs' $90 million seed and Warhound's $200 million one, all announced this week. The segment matters as much as the stage: Crunchbase counted global fintech seed funding down 35.2% from a year earlier in the third quarter, while Latin American seed rose 62%. The competition for founders has moved earlier, and it is uneven. A founder's terms now depend less on the headline market than on who, in their own segment, writes first checks and who can write the second.
PitchBook reported on October 18 that investors have funded an estimated 8,625 first-time VC financings in the US through the third quarter, above the 2021 record, while the number of active VCs has been declining for years. With the median pre-seed pre-money tripled to $12 million since 2020 and the typical seed priced at $19.5 million, investors are moving before the company exists: Brightmind Partners incubates companies for free, and even an established firm like Base10 is weighing bigger checks for smaller ownership and uncapped SAFEs. The headline says seed is booming. The real signal is that the competition for founders now starts earlier, and the investors doing it are changing their terms to win.
Dream Labs announced a $90 million seed on October 19, co-led by B Capital and Kindred Ventures, with A.Capital, Unusual Ventures, Wing VC and SV Angel participating. Founded by former Nvidia researchers, it builds Dream-1, a robot foundation model that learns skills as video prediction, offered through a public API to more than 20 hardware companies. A round this size is labelled seed but sized and staffed like a later one, and it is counted in the same seed totals as a first angel check.
Warhound launched on October 23 with a $200 million seed round and no named lead; backers include In-Q-Tel, Sequoia, ARK, Franklin Templeton, 8VC and Lux. Conceived by Palmer Luckey and Colossal's Ben Lamm, the Texas company combines Colossal's genetics with Anduril's software to supply military working dogs. No valuation was disclosed. With Dream Labs and Enact, this week's named early rounds ran from $30 million to $200 million, under the same stage labels as a first angel check.
Swiss robotics startup Enact Intelligence announced a $30 million pre-seed on October 22, backed by venture firms redalpine, Notion Capital and firstminute capital, with angels including Yann LeCun, Mistral's Arthur Mensch and Hugging Face's Thomas Wolf; no lead was named. It builds models that control different robot bodies, plus actuators that sense force and touch. A $30 million round at pre-seed, split across several funds, shows how much capital now arrives at the very first check.
Crunchbase counted $10.6 billion of global fintech funding in the third quarter, down 21.4% from a year earlier, across 687 rounds, down 31%. Angel and seed funding fell to $1.12 billion, down 35.2% year over year. The datasets differ in geography and measure: PitchBook estimates a US year-to-date record in first-time financings through the third quarter, while Crunchbase shows global fintech angel and seed dollars down by about a third in that quarter.
Crunchbase counted $1.3 billion of Latin American startup funding in the third quarter, up 26% from a year earlier. Angel and seed rounds reached $266 million, up 62%, while early-stage funding fell 34% to $288 million. Seed dollars rose in the same quarter early-stage dollars fell, but Crunchbase attributes much of the seed rise to two large rounds. One quarter of dollar data, not a count of investors: context, not a measure of who raises next.
Jim Keller's Tenstorrent closed a $1.5 billion Series E at a $6.2 billion post-money valuation on October 19, co-led by Trace Capital and Innovation X, with Fidelity, ARK, T. Rowe Price and Tether among the participants. The company builds RISC-V AI processors and servers and licenses its designs; its last valuation, in 2024, was $2.6 billion. The round also included strategic investors.
PitchBook reported on October 20 that inference-chip startups raised $4.18 billion across 22 deals in 2026 through August 28, still below 2025's record of $5.21 billion, while the median pre-money valuation reached $1.77 billion, up from $194 million in 2024. That 2026 total covers eight months against 2025's twelve; at its pace the year would pass the record. PitchBook reports the valuation median separately and does not give its sample size, so read it as a direction, not a benchmark.
Walmart chief executive John Furner announced an Agent Gateway on October 21, a server through which approved AI agents can shop, with agent profiles that set per-purchase approval or spending caps rolling out next month and agents that break its terms blocked. Amazon has been blocking agent purchases, and earlier this month JPMorgan restricted third-party agents' access to customer accounts. The rules for agents are being written by the platforms, one at a time.
OpenAI lowered the valuation it is seeking to $1.25 trillion from $1.4 trillion after some investors resisted the price, the Financial Times reported on October 22, for a raise of up to $30 billion meant as a bridge to a public offering; Coatue and UAE funds are reportedly in talks to lead, with commitments due October 30. Nothing has closed. When some buyers balk, the price moves: even at the top, a valuation is a negotiation until the money commits.
The Department of Homeland Security published an order on October 20 requiring federal civilian agencies to remove DeepSeek products immediately, including third-party products built with DeepSeek technology; it is not a ban on private or consumer use. Two days later DeepSeek closed about $15 billion at a valuation of about $74 billion, according to the Wall Street Journal and Bloomberg, and is preparing a Shanghai listing. The model a startup builds on is now a procurement question, not only a technical one.
Barclays said with its third-quarter results on October 22 that it plans about 5,000 job cuts, roughly 5% of staff, by the end of 2026, with about 1,000 already made and around 60% in support roles, and it doubled its restructuring budget to about £600 million. Chief executive C.S. Venkatakrishnan described AI as "the icing on the cake, not the cake." Talent does not vanish when large employers reorganize; it gets redistributed.
A record estimated count of first checks looks like an easier seed market. I read it as one where the competition moved earlier, often to before a company exists.
My #033 ↗ issue carried PitchBook's estimated 8,625 first-time financings. This week PitchBook reported how investors are responding: smaller funds courting founders who haven't founded anything, and even an established firm like Base10 weighing bigger checks for smaller stakes. The counterpoint: some of the price rise may reflect what is being funded, AI-heavy companies raising larger first rounds.
That gives founders leverage, not just access. If you have a choice of leads, ask each what it holds for the next round and choose on the answer. If you don't, raise enough to reach a milestone an outside lead would pay for; when the second check never comes, early employees holding options are among those exposed.
On my #033 wager: Anthropic's public registration statement, filed after this issue's window closed, reports second-quarter revenue of $11.56 billion, within 0.4% of the roughly $11.6 billion attendees said investors were told. The third quarter is blank. One quarter held; I will score the other when the filing does.
What would change my mind: if PitchBook's fourth-quarter data, due in January, puts the US pre-seed median pre-money valuation below $10 million, down from the $12 million it reports now, prices no longer support my reading of the leverage founders hold, and I will say so here. A median at or above $12 million fits my reading; it does not prove it.
Use the leverage while you have it: choose the first check that makes sense at the second. Belief becomes capital; the founder decides where it compounds.
Three passages were narrowed after publication. The January test above now says what each outcome would and would not show. The line on early employees now says they are among those exposed, not that they pay first. The Latin American seed card now notes that Crunchbase attributes much of the seed rise to two large rounds, so it reads as context rather than a measure of who raises next.