Fundamental Truths in VC today (as I see them) 1/ The barbell between large and small firms is as wide as it's ever been. These are genuinel…

Fundamental Truths in VC today (as I see them) 1/ The barbell between large and small firms is as wide as it's ever been. These are genuinely two distinct asset classes now, and LPs should treat them that way. 2/There is no portfolio construction model, whether concentrated or diversified, that rescues a bad portfolio. Construction can extend the right tail and reduce left tail risk, but GPs still have to be in great companies. What the right model looks like depends entirely on where you operate (seed vs. early vs. growth) and what your strengths are (sourcing, access, expertise). 3/Some SPV operators raising capital into the top 5 consensus growth stage names will make more in fees over the next five years than most smaller seed GPs will make in fees/ carry. I ran into someone who has deployed ~1B at 5% one-time fees and 15% carry -- In the last 12 months. 4/ The valuation gap between AI and non-AI companies is widening fast, and it's most visible at Series C, when category winners start to emerge, and large capital concentrates quickly. 5/From 2022 to 2024, growth VC was ignored. That was a mistake. Prices had come way down, and AI was in its early innings. Now the LP pendulum has swung entirely to growth, but concentrated into consensus names. Given how much these companies can raise, it's not unusual to see a large fund with only 5 to 8 positions. 6/The mid to large end of VC is increasingly taking cues from PE: secondaries, multi-product platforms including PE and credit, and heavy reliance on the wealth channel. The lines are officially blurred. 7/There is a lot of gamesmanship in ARR reporting. Founders are trained to show growth to maintain expectations and valuations. This is getting dangerous and will not end well for many of these companies. 8/No one actually knows how the next few years play out with AI. The honest truth: foundation model companies still need to find a path to sustainable margin, application layer companies are watching moats erode faster than they can build them, and the infrastructure buildout assumes demand that is still largely theoretical. The people who sound most certain are usually the most invested in a particular outcome being true. 9/Investing at 50M to 100M pre-money at seed only works mathematically if you assume you're getting into companies that exit at 10B or more. That's not impossible, but it's not easier just because everything feels hot. We saw exactly this loop play out in 2021. LPs should be asking harder questions about whether the seed funds they're backing are optimized for outcomes (3x or higher).

Source: Yohei Nakajima (X) | 2026-04-16

Related