VCs abandon old rules for a ‘funky time’ of investing in AI startups (techcrunch.com)

At TechCrunch Disrupt 2025, leading VCs including Aileen Lee (Cowboy Ventures) and Jon McNeill (DVx Ventures) said the old rules no longer apply to AI investing: rapid, sometimes explosive growth is possible, but investors are using a new “algorithm” of criteria rather than relying solely on revenue velocity. Panelists stressed that beyond headline ARR jumps, VCs now evaluate whether startups generate proprietary data, possess a defensible technical moat, demonstrate founder track records, and show real technical depth—while also scrutinizing go‑to‑market execution and customer retention. The result: Series A firms are applying more mature, granular standards earlier in the funnel, making follow‑on funding harder even for fast initial growers. The practical implication for AI/ML teams is clear: success requires a balanced playbook. Startups must build pipeline data and model improvements that create defensibility, ship features at the cadence of leading labs (OpenAI, Anthropic), and pair strong engineering with disciplined GTM to prove customer traction. Debate remains over whether GTM can overcome mediocre tech, but panelists agreed both product quality and rapid, high‑quality shipping matter. Because there are “no clear, outright winners” in LLMs and adjacent spaces, technically savvy challengers that combine deep models, unique data, and compelling go‑to‑market strategies still have realistic paths to unseat incumbents.