Martin Casado

Leads a16z's infrastructure practice and is one of Silicon Valley's most technically credible VCs, having built VMware NSX and pioneered software-defined networking. Invests 'from markets in' rather than 'from companies out' — he maps massive market opportunities first, then finds founders who can serve them. Believes AI is analogous to 1996 of the internet boom with years of growth still ahead. Sees the AI coding market alone as a potential $3 trillion opportunity. Vocal advocate for open source in AI — co-authored a piece in The Economist with Ion Stoica arguing open source is critical to AI's future. Deeply concerned about Chinese dominance in open-source AI models (claims ~80% of his portfolio companies use Chinese open-source models). Thinks AI value will ultimately accrue at the application layer more than the model layer, drawing parallels to how SaaS-era startups captured more value at the application layer than infrastructure. Believes AI is fundamentally changing SaaS pricing from per-seat to outcome-based models. Champions spatial intelligence as AI's next frontier (deeply involved with World Labs / Fei-Fei Li).

Technical founders building infrastructure that other companies build on top of. Open source projects with strong community adoption and clear monetization paths. AI companies with genuine technical moats, not just API wrappers. Developer tools that fundamentally change workflows. Companies where AI creates 4-5 orders of magnitude cost reduction over incumbent approaches. Spatial intelligence and world models. Software-defined approaches to traditionally hardware problems (inspired by his SDN background).

Companies where the 'AI' is just calling someone else's API. Infrastructure without developer love. Closed-source tools competing against strong open source alternatives without clear differentiation. Consensus deals where every fund is piling in without independent technical conviction. Companies without a clear path from model capability to defensible product.

{"education":"PhD and MS in Computer Science from Stanford University (2007). Thesis: 'Architectural Support for Security Management in Enterprise Networks.' Advisors: Nick McKeown, Scott Shenker, and Dan Boneh.","pre_stanford":"Researcher at Lawrence Livermore National Laboratory (1998-2003) working on large-scale computer simulations for the US Department of Defense. Worked with government agencies on classified network security issues.","nicira":"Co-founded Nicira Networks in 2007 with Stanford advisors Nick McKeown and Scott Shenker. Served as CTO. Built network virtualization product (shipped 2010). Acquired by VMware for $1.26B in July 2012.","openflow_and_sdn":"Created OpenFlow, the open-source protocol that enabled software-defined networking (SDN). Launched the entire SDN movement, fundamentally changing how enterprise networks are built and managed.","vmware":"After Nicira acquisition, served as VMware Fellow, CTO for Networking and Security, and GM of the Networking and Security Business Unit. Built VMware NSX from the combination of VMware vCloud Networking Security and Nicira's platform.","a16z":"Left VMware in February 2016 to join Andreessen Horowitz as its ninth general partner. Now leads the firm's infrastructure practice and its $1.25B+ infrastructure fund.","how_background_shapes_investing":"His experience building Nicira/NSX from PhD research to $1.26B acquisition gives him rare founder empathy and technical depth. He evaluates system architecture, not pitch decks. His SDN background — making networks programmable via software — directly informs his attraction to software-defined approaches in other domains (e.g., Astranis's software-defined satellite antenna). His government security background shapes his views on AI national security risks."}