Will Open-Source Threaten Anthropic's Business & Do Margins Matter in a World of AI | Matt Murphy

Anthropic Investment Strategy & Execution

• Matt Murphy credits Anjane (ex-Kleiner Perkins) with introducing him to Anthropic founders Dario Amodei and Tom Schmidt.
• Menlo invested a "starter check" of slightly over $10 million at a $4 billion valuation in the first round.
• A subsequent $500+ million Special Purpose Vehicle (SPV) was launched to lead the next round, marking Menlo's first SPV and largest single investment.
• The firm pivoted its entire strategy to be "all in" on AI despite the high valuation and lack of historical precedent for such early-stage, pre-revenue spending in a venture fund.
• Key conviction points included Anthropic's superior efficiency (spending ~1/50th of OpenAI's capital for comparable benchmark performance) and the team's technical depth.
• Murphy notes that rigid ownership mandates would have prevented the deal; Menlo prioritized access to the "biggest outlier" over maintaining a specific percentage stake.

Market Dynamics: Dilution, Valuations, and SPVs

• Murphy argues that high dilution is now the norm, even for non-frontier AI companies, driven by aggressive capital needs for offense and employee retention via secondary liquidity.
• The "winner-take-all" dynamic has shifted; owning a small percentage of a multi-billion dollar outlier is more lucrative than owning a large percentage of a $300 million exit.
• SPVs have become essential for Menlo to write checks larger than a fund's single-position mandate while keeping the main vehicle size manageable.
• Menlo avoids taking profits on winners early unless absolutely necessary, preferring to compound capital in leading companies like Anthropic and Lovable.
• Recent industry friction regarding "secondary market SPVs" prompted Dario Amodei to restrict access, curbing bad actors and ensuring only direct partners could syndicate capital for Anthropic.

Portfolio Highlights & Sector Analysis

• Lovable: Invested at a ~$150 million post-money valuation; the company grew from near-zero to ~$300 million ARR in one year, driven by "no-code" adoption.
• Legora: Invested in a sub-$50 million check six months prior; focuses on complex legal workflows with high barriers to entry, making it less susceptible to direct competition from general foundation models.
• OpenRouter: Described as a "beast" and "wildly profitable," this company serves as the "North Star" for AI model routing and optimization, intercepting API calls to select the best model based on cost, latency, and performance.
• Nebius & Chip Strategy: Murphy disputes Nebius's value in the routing business, citing OpenRouter's established developer trust, but acknowledges that companies like Anthropic and DeepSeek may move "full stack" into chip design to optimize specific workloads and reduce costs.
• Open Source vs. Frontier Models: Murphy rejects the idea that open source will displace frontier models (like Anthropic's) for 96% of enterprise use, arguing that superior performance drives higher customer retention and revenue, justifying the premium.

VC Firm Strategy: Barbell Approach & Fund Structure

• Menlo employs a "barbell strategy": pursuing hyper-growth outliers at the very early seed stage (expanding check sizes to $8M) while maintaining large growth positions, effectively skipping the congested Series A range ($1M–$3M revenue).
• The firm maintained a $3 billion fund size (rather than $10B+) to preserve a "small and mighty" culture, avoiding the misalignment risks of massive, multi-team institutions.
• Seed funds under $100 million are predicted to be the worst-performing category due to an inability to lead rounds or compete with full-stack giants (e.g., Sequoia, Founders Fund) that dominate early capital.
• Menlo has shifted away from the "negative signaling" of institutional involvement, embracing syndication and collaborative rounds as the norm.

Geography & Talent Trends

• San Francisco is regaining dominance as the central hub for AI talent, drawing back "cold" tech workers from New York and driving a renaissance in the Bay Area.
• Menlo is increasing focus on Europe, noting that founders with "grit" who succeed in the European market (e.g., Lovable, Legora) possess a distinct advantage for global scaling.
• The firm sees the "DeepMind diaspora" as a key source of emerging European talent and innovation.

Lessons Learned & Future Outlook

• Post-Deal Reflection: Murphy admits that previously "existential" losses (e.g., Plaid) no longer define career success; the focus is now on trend identification and execution over single deal outcomes.
• Overheated Sectors: The "Neo Lab" sector (60+ emerging foundation model companies) is identified as overhyped and prone to consolidation; generic labs are expected to fail.
• Under-invested Areas: The AI infrastructure stack (observability, developer tooling, model routing) is currently undervalued but critical as the ecosystem matures.
• 10-Year Vision: The most significant anticipated breakthroughs are in medical health (treating chronic conditions, drug discovery) and the fundamental transformation of the global medical system.
• Firm Sustainability: Murphy believes that post-success, the "challenger mentality" and lack of downside anxiety (due to financial security) will drive Menlo to focus on maximizing upside rather than risk mitigation.