The guy who wrote a $14 billion check before most people had heard of Claude just told us where the *next* $14 billion is hiding.
The Summary
- Matt Murphy at Menlo Ventures was an early Anthropic backer; his firm's stake is now worth ~$14 billion ahead of the company's pending IPO
- Murphy discusses how mega-rounds for foundation models are reshaping the entire funding landscape for AI startups
- He signals where smart capital is moving next in the AI stack, beyond foundation models
The Signal
Matt Murphy made one of those bets that venture capitalists dream about: early money into Anthropic, before the Claude chatbot became a household name, before the company commanded the kind of valuation that makes grown LPs weep with joy. Menlo Ventures' stake is now estimated at $14 billion. That's not a typo.
But the more interesting story isn't the win itself. It's what Murphy sees happening now that billions are concentrating in a handful of foundation model companies. The capital dynamics have changed. When Anthropic, OpenAI, and a few others are hoovering up multi-billion dollar rounds, it creates a bifurcated market.
"The money pouring into these major AI companies is impacting the broader funding environment."
On one side: the foundation model oligopoly, where checks have more zeros than most Series A rounds used to have companies. On the other: everything else. The application layer. The tooling. The infrastructure that makes AI actually useful for businesses that aren't trying to build the next GPT.
Murphy is hunting in that second territory now. He's looking at companies building *on top of* the models, not competing with them. The picks-and-shovels play is over. We're in the "what do you build with picks and shovels" phase.
Here's what matters for founders:
- If you're building another foundation model without $10B and a decade, you're late
- If you're building agents, vertical AI tools, or novel applications, the door is wide open
- The deployment gap between "AI exists" and "AI works in my business" is still enormous
The Implication
The Anthropic IPO will mark a line in the sand. Before: foundation models were a bet. After: they're infrastructure. The action moves up the stack to the companies that help businesses actually deploy these things. If you're raising in 2027, your pitch better answer one question clearly: why can't someone just build this with Claude or GPT in a weekend? If your moat is "we fine-tuned a model," you're already behind. If your moat is industry-specific deployment, integration complexity, or workflow automation that requires domain expertise, Murphy and every other smart check-writer is listening.