Michigan CS, MIT Sloan, Bloomberg backend, Microsoft cloud, Percolate product, seven years running the product organization at Cockroach Labs. And now, the operational data warehouse in Lower Manhattan.
Nate Stewart is the CEO of Materialize, a New York streaming database company headquartered on Lafayette Street, and he got the job the interesting way. He did not answer a recruiter. He did not tweet his way in. He was already on the board. In April 2024, the founder, Arjun Narayan, moved to VP of Engineering, and Stewart moved into the corner office. It is a maneuver that reads, in retrospect, as inevitable, because Stewart has spent most of a decade orbiting the same set of people.
Consider the sequence. Stewart joined Cockroach Labs in 2017, at a stage when the company had not yet begun making money, as the head of product. Two of his colleagues there were Arjun Narayan and Nikhil Benesch, the eventual Materialize founders. In 2019 they left to build a streaming database on top of an academic paper about differential dataflow. In 2020, Cockroach Labs named Stewart to its board, the first non-founder and non-investor to sit there. By 2023 he had joined Materialize's board as an independent director as well, which is how a person ends up serving on the boards of two competing-ish database companies simultaneously, and how, when the CEO seat opens, there is a shortlist of exactly one.
The path before Cockroach is the sort of thing you can trace with a public records search and a couple of MIT press releases. Stewart earned a BS in computer science, with honors, from the University of Michigan. He then worked as a lead backend engineer at Bloomberg L.P., which is either the last honest apprenticeship in software or a large, quiet financial data monastery, depending on who you ask. From there he did an MBA at MIT Sloan, specializing in entrepreneurship and innovation, and won the MIT $100K Entrepreneurship Competition. This is a thing MBAs win about once a year and then either commercialize or list on a LinkedIn profile in perpetuity. Stewart went to Microsoft afterward, worked on Cloud + Enterprise product, and then joined Percolate, a marketing SaaS platform that was later scooped up by Seismic. Then Cockroach Labs came calling.
You can read the resume as a portfolio of infrastructure that other software depends on. Financial data feeds. Cloud tools. Marketing pipelines. Distributed SQL. And now a streaming operational data warehouse that promises SQL views that stay incrementally up-to-date as new events arrive. There is a coherence to it that most careers do not have. When Stewart announced he was joining Materialize in April 2024, his line was: "The systems we interact with should reflect our dynamic world as it is in the moment, not as it was in the past." Which is, if you squint, a sentence about databases. It is also a sentence about how a person builds a career: by staying near the thing that keeps updating.
The timeline below is short but useful. Note the way the middle three roles stack: Microsoft (cloud), Percolate (marketing SaaS), Cockroach Labs (distributed SQL). Each one is closer to the machine than the last. Then a database company hires him to run it.
Approximate. Materialize tenure ongoing.
Materialize describes itself as an operational data warehouse, which is a category label the company more or less had to invent because "streaming database" was crowded and "real-time data platform" was tired. The pitch is technical but concrete: you write SQL, Materialize keeps the results current as new events arrive, and you can subscribe to changes. It is built on top of differential dataflow and timely dataflow, which are the sort of academic ideas Frank McSherry and collaborators published in the 2010s and which are now, roughly, the engine of the thing.
Stewart's job is to translate that into something a chief data officer will buy. Increasingly the phrase he uses in public is "the live data layer for apps and AI agents." Materialize is pitching itself as the memory a language model needs when it wants to know what happened in the last 200 milliseconds, not the last day. The version numbers roll forward, the customer logos accrete, and Stewart posts about it on LinkedIn every few weeks, calmly.
Stewart's 2020 appointment to the Cockroach Labs board was, by the company's own accounting, its first ever non-founder, non-investor board seat. This is unusual. Boards usually add investors, then customers, then, much later, operators. Cockroach added an operator first.
For roughly a year, from 2023 to early 2024, Stewart sat on the boards of both Cockroach Labs and Materialize. This is legal, unusual, and, to the extent both companies think of the other as adjacent, load-bearing. Then he took the Materialize job.
The MIT $100K is Sloan's flagship entrepreneurship competition. Alumni include the founders of Akamai. Stewart won it, cashed the metaphorical check, and, notably, did not go start a company right after. He went to Microsoft.
A few things about Stewart come through if you read his public output carefully. He is not a heavy poster. His Twitter/X handle exists and is quiet. His Medium account has a handful of essays. His LinkedIn is where the actual signal is, and even there, the tone is closer to product changelog than founder-influencer.
He also does the thing where operators who used to be engineers still talk in engineer's terms. When he explains what Materialize does, it is in terms of joins, freshness, incremental view maintenance, and subscription semantics, not in terms of platforms and paradigms. This is either endearing or annoying, depending on your role.
One more note. Both databases he has helped grow, Cockroach and Materialize, are cases where the engineering is famously hard, the marketing is famously understated, and the customer base is famously loyal. This appears to be a taste, not a coincidence.
Stewart's public thesis is that AI systems will not be useful for very long if the data they read is stale. Language models and agents will hit an increasingly firm ceiling if the world they are answering about is 24 hours old. The company's job, as he frames it, is to move that ceiling. Materialize is the live memory. He would like it to be inevitable.
The near-term goals, as far as they can be inferred from company announcements and his own posts, are: scale the customer base past the current Series C headcount, publish more releases aimed at AI and agent workloads, and continue the slow, patient work of turning "operational data warehouse" from a phrase into a category.
The CEO of Materialize, a New York streaming operational data warehouse. He took the role in April 2024.
Seven years at Cockroach Labs, most recently as Chief Product Officer, plus a board seat. Before that, product roles at Percolate and Microsoft, and lead backend engineering at Bloomberg L.P.
A BS in Computer Science (Honors) from the University of Michigan and an MBA from MIT Sloan, where he won the MIT $100K Entrepreneurship Competition.
He was an independent director on the Materialize board, joining in 2023. In April 2024 the founding CEO, Arjun Narayan, moved to VP of Engineering, and Stewart took the seat.
An operational data warehouse built on differential dataflow. You write SQL, Materialize keeps the answers current as new events arrive, and applications and AI agents subscribe to the results.