I sat down with a longtime LP this week. The conversation led me back to a letter I wrote to our LPs in April 2025\.
In that letter, sent as the tariff news landed and capital markets cratered, the tone was deliberately unsentimental. I feared the potential demise of true early-stage venture capital. I said the spoils would go to those with the deepest pockets, not necessarily to those with the deepest passion for the work. I said innovation would suffer, returns would suffer, and that pre-seed investing as I had practiced it for two decades was at genuine risk of becoming a nostalgic story.
I did not want to be right about any of it. I was.
Why this matters for founders: If you are a repeat founder with AI pedigree raising into the consensus story, the structural forces I am about to describe are not your problem. You can get $10M from a16z; the data says you may not even need to ask. But if you are a first-time founder, a non-consensus technical founder, or a founder in a category that the mega-funds have not yet noticed, the capital available to you at inception is shrinking. The contraction is real, and it is bifurcating who gets funded at the earliest stages.
What the data confirms
The April 2025 worry was structural, not cyclical. The year since has put numbers on what I could only describe as a fear.
Twelve firms collected more than half of all U.S. venture cash in the first half of 2025\. Founders Fund alone raised $4.6 billion, more than double what all 44 first-time fund managers raised combined in the same period. Total U.S. VC fundraising came in at $66.1 billion for 2025, the lowest figure since 2017\. The number of first-time U.S. funds: 77 through mid-year, down from 215 in all of 2023\. That is roughly a 64% contraction in the emerging-manager bench, the sharpest on record.
One allocator’s framing has stuck with me. He told me it is “…hard to convince (us) to invest in a $50M fund unless you’re super pedigreed: maybe if you’re the co-founder of OpenAI.” Two decades of experience investing at inception, apparently, is no longer the credential. Pedigree is. That framing has since become policy: endowment participation in emerging-manager funds dropped 18 percentage points over the past year; fund-of-funds participation dropped 21 points. The institutional emerging-manager basket I called for in April 2025 is moving the wrong direction.
Meanwhile, median seed valuations hit approximately $20 million in 2025, up 35% year-over-year. AI startups commanded a significant premium above that. I wrote that commanding 10% ownership in a diverse enough basket of technical positions now requires a $100 million pre-seed fund. The math has not improved: 48% of 2025-vintage U.S. VC funds were sized between $10 million and $100 million, the smallest share in nine years. The sub-$50 million conviction-first model, the one Bee was built on, is being priced out structurally.
Why this matters for founders: If the firms that historically wrote the earliest checks are contracting or disappearing, the capital available at inception shrinks. You are not imagining a harder fundraising environment at the pre-seed stage. You are experiencing a documented structural contraction.
And then… SaaSpocalypse 2026
The April 2025 thesis was about venture market structure: capital concentration, exit drought, LP consolidation. What happened in late 2025 and early 2026 is a separate, compounding event.
Agentic AI capabilities crossed a threshold in late 2025 and early 2026\. The market responded on February 3, 2026: approximately $285 billion in SaaS market cap erased in 48 hours. Total software-sector losses extended past $1 trillion by early March. Industry coverage is treating it not as a dip but as a structural unraveling of SaaS moats. Subscription revenue from software that AI can now replicate or substitute has been repriced.
This matters for the venture ecosystem in one specific way: the IPO pipeline for SaaS-model companies, already frozen from the exit drought, may now face a permanently altered public market, at least until they redirect their offerings to be AI-enabled. The companies that would have been the 2025 and 2026 liquidity events for LPs and GPs are entering a market that is re-evaluating their business model category in real time. That freeze has consequences upstream: fewer distributions, more hesitant LPs, more manager attrition.
I am not predicting where SaaS valuations settle but instead making the point that the structural problem I described in April 2025 now has a compounding layer that was not even visible then.
And even as the contraction unfolds, the spoils accrue upward. One firm raised $15 billion in 2025 alone, more than the next two largest fundraises combined, roughly 18% of all U.S. venture capital that year. The mega-funds are not weathering the consolidation; they are the consolidation.
So, what can we do about it?
I ended my April 2025 update with a commitment: I would advocate for more capital commitments to sub-$100 million funds, for institutions to create emerging-manager baskets, and for GPs to collaborate far more, period. The case is even stronger now, not weaker. Let me say it precisely.
Sub-$100 million funds are where venture math still works, but only when they are hyper-specific. The era of the generalist sub-$50 million pre-seed fund is long over. Median seed valuations are $20 million post-money; anything less is treated as adverse selection. At that floor, the only sub-$100 million managers who can still generate fund-returning outcomes are the ones with a defined edge: a single sector they know better than anyone, a community connected by a common bond (a university, a geography, a technical discipline), or a singular “future” they believe in deeply enough to bet a whole portfolio on. Specificity is what lets a sub-$100 million fund find the companies the mega-funds can’t see and still command ownership the math respects.
When LPs consolidate entirely toward platform funds for “reliability and persistence,” they are not reducing risk. They are concentrating exposure to the same deals, the same stages, and the same pricing that every other mega-fund is chasing. The asymmetric return potential that made venture the asset class comes from pre-consensus, pre-obvious, pre-revenue conviction. That conviction does not live at scale. It never has.
Why this matters for founders: The sub-$100 million funds that get funded by LPs become the firms that fund you. The contraction happening at the fund-formation level now will reduce the supply of genuine pre-seed capital available to founders in 2026, 2027, and 2028\. Not because VCs are afraid. Because their LPs have stopped backing them.
Institutional emerging-manager baskets are alpha. The firms that will generate the next generation of venture returns are relatively unknown. They are building their track records now, in the hardest fundraising environment in a decade, with the smallest institutional support in memory. The endowments and foundations that stepped back from emerging managers over the past year are optimizing for the appearance of safety at the cost of actual return generation. A dedicated basket, sized appropriately as a percentage of a venture allocation, accesses the part of the market where non-consensus bets still live. Ownership has compressed since prior cycles; entry prices have not yet been bid up beyond the reach of asymmetric returns. A diversified basket of these hyper-focused emerging managers is also how an LP captures the asymmetric upside while diversifying away the sector-timing risk any single fund carries. The hyper-focused GP who gets timing wrong gets wrecked. The LP with twenty such managers across as many sectors does not. The basket structure transfers the riskiest bet (being right about which sector hits next) from the LP to the GP, where it belongs.
Emerging-manager GPs need to collaborate, period. The solo-GP pre-seed model, which (now) describes my firm and many of the managers I respect most, was built on the assumption that differentiation and track record would carry on their own merits. That assumption held in a functioning market. It does not hold in a market where twelve firms are collecting half of all capital. Collaboration does not mean combining funds or giving up independence. It means co-investing where conviction overlaps, sharing intelligence about portfolio companies and stages, jointly advocating for the institutional structures the sector needs, and convening LPs as a category rather than as individual managers each shouting into the same room. The kind of weight that none of us can carry alone becomes possible when carried together.
Where Bee stands
I have been a pre-seed investor for two decades. I have backed nearly 100 technical companies at inception across AI, robotics, advanced manufacturing, and synthetic biology. More than 60% of those companies historically reached Series A. The portfolio has raised over $2 billion in follow-on capital. I have lived through multiple cycles, and I have never wavered on the stage or the strategy.
The current environment is real, and the structural forces I described twelve months ago have tightened, not loosened. What the last year has done is confirm the analysis and sharpen the prescription: back managers who are doing the work of identifying founders before the market sees them, at prices where the math works, with the kind of partnership that actually helps companies compound. Make sure those managers have access to capital. Create the institutional structures that protect the part of the venture ecosystem where real innovation is seeded.
Why this matters for founders: The fund manager most likely to write your first check, if you are building outside the consensus, is a sub-$100 million emerging manager with a defined edge. Whether that manager exists in 2027 depends on whether their LPs back them in 2026\. Your inception capital is being decided right now, in conversations you are not in.
Here is where I am: excited about the innovation happening at inception right now; but less so about the cycle making it harder to fund. And, I’m hungry for collaboration with the LPs and GPs who still believe early-stage returns come from sub-$100 million fund managers. And I’m still writing checks.
If you are an LP who shares this perspective, or a GP keen on working together, email me. And founders: we are active. Let’s be partners.