Michael Dempsey

Who they are

Michael Dempsey is General Partner at Compound — known for the essay 'Narratives & Pseudosecrets' on being early vs. wrong in tech, and writes at michaeldempsey.me on deep tech and value capture.

Person

Michael started at CB Insights as one of the first 15 employees, helping build the data and research function when the firm was still early-stage — a front-row seat to how information asymmetry shapes venture decisions. From there he moved into investing, eventually becoming General Partner at Compound, a firm built around long-horizon bets on deep tech. He writes prolifically across michaeldempsey.me, a Substack, and Medium — his essays cover AI, robotics, autonomous vehicles, and the structural problem of being too early to a technology wave. His 2019 piece 'Narratives & Pseudosecrets' is a sharp take on how tech narratives travel and why being early is often indistinguishable from being wrong. He has appeared on podcasts and in interview series — Infinite Frontiers, The Inquisitive VC, and a 2025 episode titled 'Why Deep Tech Startups Mess up Value Capture' — and his LinkedIn series 'On My Mind' covers AI dynamics and what he calls 'meta-moats.' The through-line is a consistent obsession with timing, information edges, and where value actually accrues in frontier technology.

Network

The available signals show Michael engaging with the broader venture and media community rather than a tight named inner circle — he has been featured by Nawaz Ahmed at The Inquisitive VC and profiled in the Moth Fund interview series. No direct co-investor or portfolio-company relationships surface from the current claims.

Nawaz Ahmed· Host, The Inquisitive VC

How they likely show up

Active writing across a personal blog, Substack, Medium, and LinkedIn → he processes ideas publicly and iteratively; he's likely comfortable debating half-formed theses out loud.
Recurring focus on timing risk ('being early vs. being wrong') and value capture in deep tech → he will stress-test assumptions about when a market is actually ready, not just whether it will exist.
Moved from data/research at CB Insights into investing → he is oriented toward information infrastructure and probably values rigorous sourcing over narrative alone.
Podcast topic in 2025 was specifically 'Why Deep Tech Startups Mess up Value Capture' → he thinks structurally about where returns land in a stack, not just whether a technology works.
Possibly — medium tenure shape at Compound suggests he has been at the firm long enough to have a developed point of view on the portfolio but may be looking to sharpen the next fund's thesis.

Conversation tips

→ Reference a specific essay — 'Narratives & Pseudosecrets' or the 'Meta-Moats of AI' post — so he knows you've read the actual work, not just skimmed a bio.
→ Engage on timing risk: ask where he thinks the market is relative to where a technology actually is — that's the frame he returns to repeatedly.
→ Don't pitch a deep tech company as 'ready now' without a crisp answer on value capture — that's precisely the failure mode he's written and spoken about most.
→ His CB Insights background means he respects data-grounded arguments; bring a specific market signal, not just a trend narrative.
→ Ask about the Substack post 'VC-Backed Startups are Low Status' — it's a pointed take and he'll appreciate you engaging with the argument rather than glossing over it.

Toolbox

Openers

Open on the 2025 podcast episode 'Why Deep Tech Startups Mess up Value Capture' — it's his most recent named public appearance and zeroes in on the structural question he cares most about: who actually captures value when a frontier technology matures.
Reference 'Narratives & Pseudosecrets' — his 2019 essay arguing that being early to a technology is functionally indistinguishable from being wrong — and ask how he thinks about that tension inside his own portfolio today.
Bring up the Substack piece 'VC-Backed Startups are Low Status' — it's a deliberately provocative frame and signals you've gone past the surface-level LinkedIn posts to engage with the argument.

Discovery questions

Your 2025 talk was specifically about deep tech startups messing up value capture — what's the most common structural mistake you see founders make when thinking about where margin lands in their stack?
You've written about the danger of being early vs. being wrong — how do you decide, inside Compound's portfolio, when to hold conviction on a timing call versus treat it as a genuine mis-read?
The 'Meta-Moats of AI' post argues that durable advantage in AI is shifting — where do you think that actually lands today, and has your view changed since you wrote it in 2024?

Avoid

Don't lead with a broad 'AI is transforming everything' framing — he has written explicitly about the difference between narrative momentum and real value capture, and a vague AI pitch will signal you haven't done the reading.

Make it yours