Things I Think I Think - Q3 2026 — Chris Neumann
Short summary
Chris Neumann's Q3 2026 take on venture capital covers the widening gap between hypergrowth AI companies and everyone else, the coming disillusionment with AI agents, and how today's most competitive founders pitch in a fundamentally different way than they did two years ago.
- VC funding is splitting into two tiers: mega-funds chasing hypergrowth AI companies at inflated valuations, while most early-stage companies face a sluggish market. Series A and B investors are now deciding whether to chase hot companies regardless of cost, stick to pre-AI valuations, or adopt a hybrid approach.
- AI agents have crossed the hype peak and are heading toward the trough of disillusionment. Users are discovering agents fail on edge cases, require more oversight than doing tasks manually, and the real moat will be customer support—expect negative media stories once agents cause financial harm to regular people.
- Competitive founders today pitch opportunities measured in tens of billions or trillions, skip traditional TAM/SAM/SOM sizing, and credibly claim 2-3 year paths to $1B revenue with multiple funding rounds per year. Founders pitching slower timelines or multiple rounds just to reach product-market fit now look out-of-touch to VCs.
Full summary
The VC market is splitting hard between companies showing AI hypergrowth and everyone else. In Q3 2026, preemptive Series A and B rounds accelerated again after a summer slowdown, but this only benefited a narrow slice of startups. Meanwhile, most companies face a brutal fundraising environment. VCs are now applying the same triage strategy that started at Pre-Seed in 2025 — either chase the obvious winners or pull funding entirely. Bridge rounds are drying up as early-stage investors abandon portfolio companies rather than throw good money after bad. The real question looming in Q4 is whether funds will chase these hypergrowth companies regardless of valuation, stick to older pricing models, try a hybrid approach, or simply exit the game. On the flip side, strong companies that would've been slam-dunk raises a year ago are hitting pause if Q4 responses disappoint, waiting for the IPO window to reopen instead.
The IPO market remains stuck. Anthropic's S-1 leaked last week with early November as the target date, but obstacles keep piling up. SpaceX's IPO landed with a whimper, other tech IPOs got pushed back, and liquidity across the board is evaporating. VCs themselves are getting squeezed — the average time between raising a first and second fund has spiked to nearly 4 years versus 2.5 years in 2022. When your investment period is 2-3 years but fundraising takes 4 years, you're sitting idle for a full year. This crunch is probably worse outside the US, where mega funds and high-profile GP spin-outs are the only ones raising quickly.
AI agents have hit the hype peak and are now crashing toward the trough of disillusionment. A sub-one-year-old startup just raised a billion, Facebook released an agent that got blocked everywhere, and Apple finally shipped a working Siri. But here's the problem: agents barely work. Users are discovering they're finicky, limited, and full of edge cases. Most people in the Valley use them only for low-stakes tasks. The author stopped using agents entirely because verifying whether they actually completed the task correctly took longer than doing it himself. Real damage is coming when mainstream users hit the gap between hype and reality — once the news starts covering stories about an agent draining someone's retirement account or a small business's payroll, the backlash will be swift and brutal. The actual moat for AI agents will turn out to be customer support, not the technology itself.