A coalition of major Silicon Valley-based AI tech platforms and VCs just published a letter on open weights and American AI leadership. NVIDIA, Meta, Microsoft, OpenAI, a16z, YCombinator, and over a dozen others signed it.
We won't wade into the open vs.closed debate – or open weights vs. open source. Those discussions belong to the people building models, not to venture funds backing founders who build on top of them.
But three statements in the letter validate exactly what Mercury Fund has been saying about Vertical AI.
First, match the right model to the right job at the right cost. Frontier capability isn't the point. Most real work in most industries doesn't need it. Vertical AI startup Collide built an agentic orchestration platform for oil and gas companies that doesn't touch frontier pricing for most of its queries. SmartAC runs diagnostics across thousands of home HVAC systems and communicates with contractors the same way. Neither platform needs a model that can write poetry. They need one that's fast, cheap, and right, over and over.
Second, own your accumulated knowledge instead of renting it. Medefy has spent years building a proprietary understanding of how employees navigate health benefits. The team behind Omniscience has spent even more time studying how pharma and biotech companies manage clinical trials, turning years of trial data into complex decision layers no outside frontier model has access to. That knowledge compounds on their side of the relationship, not a frontier lab's.
Third, competition lives at the application layer, not at the model layer. Nectar Social didn't win customers by picking abetter model. It won by understanding how brands interact with customers within DMs and then training those social media agents on each authentic brand voice. Similarly, Pie was built by two founders who know the SMB market cold. They're winning because of the domain expertise and distribution networking, and are training models on how to automate marketing for local businesses. The LLMs are infrastructure. The unique agents and industry-driven workflows are the business.
Mercury Fund has focused on B2B software investing since inception, always with an eye towards vertical orientation. Over the last decade, we've built deep relationships with and understandings of industrial ecosystems, learning what it takes to win mid-market and enterprise clients where the last mile of workflow is where value gets created. That effort has been led by our co-founder, Blair Garrou, and shaped alongside him by me and our investment partners.
As the next era of AI technology reshapes how those industries work, it feels almost poetic that the vertical software thesis we've held for over a decade is exactly the profile the Vertical Era of AI rewards.
