General Partner at Greylock Partners
Check size: $500K-$200M (sweet spot ~$25M)
Coined the 'Systems of Intelligence' framework in his landmark 2017 blog post 'The New Moats' — the idea that AI and data create competitive advantages by sitting between systems of record (databases) and systems of engagement (apps). The intelligence layer crosses multiple data sets and multiple systems of record to predict behavior, churn, LTV, or serve timely content. This layer is where value accrues and where startups build defensible moats through data network effects. Updated the thesis in 2023 with 'The New New Moats,' addressing how foundation models and LLMs change the equation: open-source LLMs shift value to apps and infrastructure around the models, but startups risk 'thin IP' if they're just wrappers around ChatGPT. Also developed the 'Unit of Value' framework — the smallest measurable unit at which your product delivers value — which determines how you price, scale, and sell. Believes good investors must be optimists, and VCs get in trouble when they move outside their 'strike zone.'
Frame your product as a 'system of intelligence' — show how you sit on top of existing data and create compounding value through data flywheels. He loves enterprise founders who understand their buyer's workflow and the enterprise sales motion from the inside. Come with architecture diagrams and data moat arguments, not just metrics. Articulate your 'Unit of Value' — what exactly does the customer pay for and how does that scale? Show open-source traction or developer community adoption if applicable. Demonstrate that your AI is core architecture, not a bolt-on feature — avoid looking like a 'thin IP' wrapper. He wants to 'earn the right' to work with exceptional founders, so show deep technical insight and obsessive focus on the problem.
Enterprise companies building 'systems of intelligence' on top of existing data. Founders with deep technical insight into how data creates compounding advantages. Products where AI gets better as more customers use them (data network effects / data flywheels). The middleware stack for LLM applications — data ingestion, indexing, and query layers. Open source projects with strong early community traction and short time-to-value. Cloud-native infrastructure that rides the permanent shift to cloud-only. Companies that can become the next $10B-$100B defining platform (the next LinkedIn, Workday, or Facebook).
Companies adding 'AI' as a feature to an otherwise unremarkable product — 'thin IP' wrappers around ChatGPT. Enterprise tools without clear data advantages or defensible moats. Founders who can't articulate their technical moat. Deals outside his strike zone of enterprise, cloud, data, and AI. Companies that are 'good but not great' — he's looking for the exceptional outlier, not the solid base hit.
“The most inspiring founders have that focus, that obsession, that belief that the world needs what they are building.”
— Greylock
“When I meet a world-class founder with the right combination of passion and grit, I'll pursue the opportunity as aggressively as I can.”
— Greylock
“Finding the right founder-investor fit can help smooth the journey. I recognize that I'm seeking to work with exceptionally talented individuals, and I want to earn the right to work with them.”
— Greylock
+ 178 more investments. View fund →