The Twenty Minute VCWill Open-Source Threaten Anthropic's Business & Do Margins Matter in a World of AI | Matt Murphy
At a glance
WHAT IT’S REALLY ABOUT
Matt Murphy on AI investing: ownership, margins, open-source, barbell strategy.
- Menlo’s Anthropic investment required abandoning traditional ownership/valuation heuristics in favor of gaining exposure to rare, compounding outliers.
- Murphy argues margins still matter in AI apps, but investors must underwrite credible paths from today’s 20–30% gross margins to ~60–70% via optimization, model strategy, and cost control.
- He believes open source will meaningfully share workloads but won’t “displace” frontier models like Anthropic where performance drives retention, revenue, and user engagement.
- The market is shifting toward multi-model optimization (routing, observability, tooling), making infrastructure layers like OpenRouter increasingly valuable as companies manage price/performance/latency tradeoffs.
- Venture insertion points have changed: Series A is unusually risky/expensive, pushing Menlo toward a barbell approach—earlier seed/neo-labs and later “anointed winner” growth rounds—while keeping fund size and culture deliberately constrained.
IDEAS WORTH REMEMBERING
5 ideasIn AI, “being in” outliers beats owning a lot of average outcomes.
Murphy says outcome sizes have expanded so much that small percentage ownership in a generational winner can outperform high ownership in a $300–500M exit; this drove Menlo to prioritize access over perfect entry price.
Price matters less than fit and exposure—until you truly can’t deploy effectively.
Rather than anchoring on valuation, Menlo focuses on whether it can play meaningfully within fund guardrails; when check size exceeds mandate, SPVs and LP co-invest can extend capacity without resizing the core funds.
SPVs are powerful, but the market’s trust problem is real.
Menlo’s first SPV (>$500M) was “nerve-wracking,” and Murphy notes founders are increasingly hostile to unauthorized SPV marketing and secondary “access” claims—cap-table control is tightening.
Margins still matter; the key is believing the path, not the snapshot.
Many AI apps look margin-light today due to inference costs, but Murphy underwrites whether teams can credibly move from ~20–30% gross margin to ~60–70% via optimization, model mix, and possibly complementary in-house models.
Open source will take some calls, but frontier performance can still win economically.
Murphy expects enterprises to mix models (e.g., premium calls to frontier, cheaper calls to open source), but argues higher-performing models can pay for themselves by improving retention, revenue, and engagement.
WORDS WORTH SAVING
5 quotesWe're in an outlier business right now, right?
— Matt Murphy
There's never going to be a perfect entry point into this market. If we wanted to be in this market, this was the way in.
— Matt Murphy
You have to be in the big outliers to drive great returns, and you're better off being in them at a very small percent than owning a large percent of a company that exits for 3 to 500.
— Matt Murphy
Wave one of AI is like, "Let's just get, let's just get it going." Wave two is like, "Let's get a lot more sophisticated about what we use and when and how."
— Matt Murphy
This company is already a beast.
— Matt Murphy
High quality AI-generated summary created from speaker-labeled transcript.
