Back to stories

Menlo’s Investment in Neon: Serverless Postgres for Modern Workloads

August 1, 2023

Our rationale for investing in Neon was remarkably clear. Menlo Ventures had been on the hunt for a modern OLTP database with separation of storage and compute, akin to Snowflake’s game-changing approach to OLAP, for quite some time.

When the Neon team first described/explained/demoed the product—it hit me like an anvil falling off a cliff. It was one of those obvious, visceral moments in life. It’s a complicated analogy to complete, but I just wanted an anvil to appear in this post. Regardless, the proverbial anvil hit me in the head, Roadrunner <> Wile E. Coyote style.

Usually, I’m the roadrunner. But, I had this exact dream a few nights ago with a Postgres roadrunner chasing me with a Neon-branded anvil. Please don’t call me old because I’m using Looney Tunes analogies.

Now that we are done with strange analogies, this article will iterate through the very clear and obvious reasons why Neon was a no-brainer Series B investment for Menlo.


Nikita Shamgunov was one of those people you’re connected to on LinkedIn—but didn’t ACTUALLY know in real life. Everyone has these connections.

Our first meeting was at Nikita’s home on a Saturday afternoon. He immediately jumped into a demo while one of his children looked over his shoulder, pressuring him to swim later. He walked me through a board deck and then kicked me out so he could swim with his kids, which impressed me. I appreciated that he prioritized the Saturday afternoon with his family and not any of the many VCs chasing him.

Over time, we bonded over a shared passion for Barry’s Bootcamp, post-workout shakes, and discussions at Stanford Shopping Center. We started taking the same classes in order to meet afterward. We soon fell in love. (OK, I made up that last part.)

Nikita’s resume is impeccable: He’s a multiple-time founder, having previously founded MemSQL and SingleStore). He is currently a partner at Khosla Ventures. He’s deeply technical, has an incredible product mind, and his grasp of the market is unlike anything I’ve seen. When you meet him, it’s immediately clear that he will make Postgres even more dominant than it already is. Moreover, Nikita will ensure Neon continues to grow to be an incredible open-core product for developers, with deep integrations/partnerships into the developer ecosystem.

Board meetings on the treadmill, especially on Zoom, tend to be challenging. (Left to right: existing investor Glenn Solomon, Nikita, Tim).

Postgres—All Open-Source

Despite being in existence for 27 years, Postgres continues to grow in influence and use. With its robust offering of varied extensions, Postgres had a bit of a renaissance over the past three to four years, to the point that it’s now, arguably, the go-to database for developers; over other popular databases such as MongoDB and MySQL. Developers are quickly consolidating their workloads across various databases into a single, multi-purpose solution. Using Postgres, developers are consolidating use cases, avoiding data drift, reducing complexity, cost, and much more., a highly-respected and referenced resource on database popularity, has several interesting charts reflecting this. As other relational databases have been primarily flat to negative, Postgres usage (orange line) has snowballed while eating into SQL Server and MongoDB.

Postgres is on a vector (no pun intended) to become the dominant DB (not just relational). You can read more about the ranking methodology here.

The Neon team loves Postgres. Their goals were not to modify Postgres, and any changes the team made were to be committed back to Postgres open source. The team has a 1,000-line patch delta to deal with the separated storage of compute and page reading/eviction, which is necessary given the Neon architecture. Neon aims to have any Postgres instance run on the Neon compute platform.

In addition to Nikita, Neon is led by several fantastic folks, including co-founders Heikki Linnakangas and Stas Kelvich. They’ve brought on a number of leaders, contributors, and committers from the Postgres community. Heikki is a longtime Postgres hacker and committer. He did an excellent talk for Andy Pavlo’s Database Seminar Series, a must-see for any Postgres or Neon fan.

Modern OLTP: Separated Storage and Compute

The team at Neon is united by a clear mission: End the burden of database management through the separation of storage and compute (this is table stakes these days), but move compute to the edge in an ephemeral manner. Through this approach, Neon frees devs to spend time on business logic and improving apps, while Neon management software deals with efficiency, scalability, capacity planning, backups, and uptime. Any management, if any, is completely done in the robust Neon UI:

Neon’s UI is better than consumer products. I wish AirBnb and TikTok were this beautiful.

Scale to Zero

The holy grail for today’s devs is scale-to-zero. By providing ephemeral compute using microVMs with scale-to-zero out of the box, Neon quickly beats other databases missing this feature; developers only pay for compute they use. As an example, the graph below shows idle time on Aurora being paid for when it’s not in use:

Copy on Write Branching

Neon created a unique concept for databases: branching. Branches in the database world are much like branches in a codebase. You create a duplicate copy (or backup, if you like) for testing, development, or otherwise. A simple yet novel feature for a database! Developers can create as many branches as they want since they’re copy-on-write. One can time travel into the past and recover from unfortunate SQL deletes/drops.

If only the NYC subway were this simple.
In real life, I use better names for tables, databases and branches. I promise.

Dev Platform Integrations

Being a developer-oriented database platform infers that developers are your main constituency and primary customer. Nikita and the team realized this. Early on, they focused on building deep integrations and partnerships with the most prominent developer platforms around, including:

Direct integration, not just, “Please set these env variables and copy this code.”

If you want to geek out, you can learn more about the Vercel/Neon integration in this great Fireship video.


One of the most potent aspects of Postgres has always been its extension ecosystem. Not only does Neon support all of the existing Postgres plugins, but the Neon team has been creating custom ones to meet the needs of their customers, including PG_Embedding, which provides an approximate nearest neighbor (ANN) query function based on HNSW. See this exhaustive Gist for a list of available and mostly uncategorized Postgres extensions.

Looks like pg_embedding (yellow) is faster. Not sure though.

Note there’s also a Langchain integration already available:


At Menlo Ventures, we are beyond excited to be a part of Nikita, Stas, and Heikki’s journey to bring Postgres to the forefront of the OLTP and developer universe. By taking an open-source-first, community/dev-prioritized approach, they will continue to win the hearts and minds of developers and build a company that will surely last the test of time. To learn more about Neon, start with their blog page and sign up for an account to get going!