Dave Munichiello of GV on investing across all stages at YC, the GitLab story (passed on Series B, backed it after), and Google as a permanent LP

Summary

Dave Munichiello, managing partner at GV (Google Ventures), has been investing in AI and enterprise software for 13 years. His conversation at YC Demo Day 2025 covers GV's structural advantages, its multi-stage approach to YC, and where he sees the next wave of AI value accruing.

The GitLab lesson

GV passed on GitLab's Series B. The concerns at the time were the remote-only model — GitLab and WordPress were among the first companies to operate that way — and skepticism about competing with GitHub. GV backed the company shortly after the Series B and remains a public market shareholder today. Munichiello stepped off the floor to join the earnings call with the CEO and CFO just before this conversation. The pass-then-back story is a clean illustration of GV's stated posture: passing on a round is not a relationship ending, it is a pause.

Permanent capital, no LP pressure

Google is GV's sole LP. Munichiello describes it as essentially permanent capital — GV has a half-hour conversation with its LP every couple of years and faces none of the quarterly deal-flow interrogation that most GPs navigate. The tradeoff is that GV does not spend time on LP management, investor days, or communications cycles, which is why Munichiello says he cleared the appearance here with his comms team. That structure frees the firm to focus entirely on founders and other GPs.

Multi-stage investing at YC

GV writes checks across seed through public markets — they cite a $100M position in a public company in the same breath as seed bets at Demo Day. Series A and B are their sweet spot. Every year they run an event with YC founders a few weeks before Demo Day and stay close to the group partners. At Demo Day itself, they have met roughly 30 to 40 companies and expect to invest in a handful.

The constraint on seed activity is signal risk. When GV does an early-stage bet, Munichiello says they apply the same criteria they would use for a Series A lead — effectively asking whether they would lead the A if the company had metrics. That discipline means GV cannot scatter capital across 30 companies the way a dedicated seed program can. General Catalyst, by contrast, runs a separate seed program that lets it categorize early bets differently and manage signal risk independently.

On Demo Day heat specifically, Munichiello is relaxed. The hottest companies at Demo Day do not reliably become the highest-revenue companies over two or three years. He notes a founder in the current batch raised nearly $8 million in uncapped convertible notes, which he frames as a sign of founder quality attracting enthusiasm — but also as a structure that is genuinely hard to evaluate given only three months of operating data.

AI as sustaining and disruptive simultaneously

Munichiello rejects the binary framing of AI as either good for incumbents or good for startups. The internet was good for established companies that adapted and for entirely novel ones; AI can do both. His portfolio examples span vertical AI — Harvey in legal, Sierra and Decagon in customer support, unnamed medical AI companies — but his more open question is what comes after single-function replacement. Replacing one human is tractable; replacing a whole functional team simultaneously, cross-functionally, is the harder and more interesting problem.

He echoes a framing attributed to Dylan Patel: consumer LLMs and code generation were the first two large AI revenue categories, and business automation is the third. The agents framing he considers somewhat simplistic — the more interesting version is autonomous systems that do not interact like a regular employee at all.

On the Mag 7, his position is straightforward: big tech went from hundreds of billions to trillions in market cap, and there is room for the landscape to expand further rather than consolidate. The 49% investment structure — a reference to deals like the Windsurf structure — he reads as the new M&A, a way to capture talent and optionality without a full acquisition.