The Engineer Who Went Long on Everything

In 2010, a former Intel software lead walked out of Stanford GSB and into the first Felicis Ventures fund meeting with $41 million to deploy. The fund was small. The bets were not. Fifteen years later, Sundeep Peechu is sitting on a $900 million fund - the firm's tenth - and the portfolio behind him looks like someone curated the defining companies of two technological eras.

The thing about Peechu is he never stopped being an engineer. He coded at Sarvega before Intel bought it. He thinks in systems, in infrastructure, in what happens when the underlying layer shifts. When he backed Plaid in the early days, he wasn't betting on a fintech app. He was betting on the plumbing that every fintech app would eventually need. Same with Supabase. Same with Weights and Biases. Same with the dozens of companies in his portfolio that most people didn't hear about until they were already massive.

The Long Game

Here is what Peechu keeps saying, in tweets and on panels and in interviews: most people misunderstand how venture returns work. They picture a quick flip, a hot IPO, a three-year payday. But when he talks about Adyen and Shopify - two companies he uses as proof of something - he points to year nine, when Adyen had $80 million in revenue and Shopify had $150 million. By most metrics, those were good but not legendary numbers. Then they compounded. Nearly all of the legendary status accrued in the second decade.

That's not a philosophy that sounds exciting at a cocktail party. It's the kind of thing that separates the funds that matter from the funds that make noise. Peechu has been at Felicis since the first external fund. He wasn't chasing headlines. He was building a body of work.

His own career has operated the same way. Born in India, he made it to IIT Madras - a school that accepts a fraction of a percent of applicants - then to the University of Illinois for a master's in computer science, then to a startup in Chicago building XML security infrastructure. His parents lived apart for over a decade so he could receive a better education. That's not trivia. It's context. When Peechu writes a check to a first-generation founder building something improbable, he's not performing solidarity. He knows the arithmetic personally.

From Sarvega to Felicis: The Intel Years That Weren't Wasted

Most venture origin stories skip the middle. Peechu's middle is instructive. After Sarvega got acquired by Intel in 2005, he stayed. Spent years in product and technical lead roles inside one of the world's biggest technology companies. He watched how enterprise software actually gets adopted - slowly, with procurement cycles and IT committees and integration nightmares. He watched how infrastructure decisions made at a big company ripple out for a decade.

Then he went to Stanford, did the MBA, and came out on the other side convinced that the founders who understand the full stack - product, distribution, infrastructure, enterprise sales dynamics - are the ones who build durable companies. That's the lens he brings to Felicis. Not just "is this product exciting?" but "does this person understand every layer of the problem they're solving?"

The AI Thesis He's Been Running Since Before It Was Fashionable

Felicis started investing in AI companies before "AI investing" was a competitive category. Runway. Poolside. Weights and Biases. Deep Infra. Flower Labs. Each investment has a logic: infrastructure before applications, tools before outcomes, the layer that everyone else will eventually depend on.

In September 2024, Peechu published a thesis that cut through a lot of noise: AI is catalyzing a long-awaited enterprise software promise. Not because AI is clever. Because it's shifting the market from SaaS, where you pay for access to software, to AI co-pilots and autopilots, where you pay for outcomes. That shift, he argued, increases market sizes by 10x or more for the startups that get it right. Traditional SaaS has a ceiling. AI-native software doesn't have one yet.

After surveying more than 40 Chief Information Security Officers, he made another call that got attention: AI security will be bigger than cloud security. Not a prediction for the distant future. A structural observation about what enterprises will need to spend money on as AI becomes critical infrastructure.

How He Actually Works With Founders

Here's the thing that doesn't show up in portfolio press releases: when Mercor was scaling fast and needed to build a team quickly, Sundeep Peechu personally took calls with almost every candidate the company was considering. Not because he had to. Because he decided that was the right use of his time at that moment.

This is what "hands-on investor" actually looks like versus the version where someone says it in a pitch meeting. It means Peechu's calendar has a lot of stuff on it that other managing partners at funds this size would delegate to associates. He connects founders pre-launch. He mediates disputes when founding teams hit friction. He recruits. He thinks of himself as one node in the network that makes companies work, not as the check-writer at the beginning and the board observer at the end.

That approach scales in weird ways. His Medium blog - titled "Hacking Venture @ Felicis" - has 1,500 followers and exists specifically to demystify how venture works for founders who didn't go to Stanford or grow up in Silicon Valley. He's explained power law math publicly, walked through fund mechanics, written about what makes a 40-year VC career - which he thinks of as a portfolio of roughly 10 high-risk bets, most of which won't compound, but which need to produce real outliers when they work.

The Health Obsession Nobody Talks About

Peechu's investments in Prenuvo and PicnicHealth and Cedar aren't random. He has a thesis about health that's more personal than most investors admit to. He believes that frequent blood testing and body optimization - the kind of thing that currently costs thousands of dollars per quarter and is mainly done by CEOs under extreme stress - will eventually become mainstream consumer technology. The results, he argues, are undeniable. The cost just needs to fall.

He backed Prenuvo (whole-body preventive MRI scanning) and Komodo Health (healthcare data infrastructure) and Recursion Pharmaceuticals (AI drug discovery) from that same frame: technology will eventually democratize access to the kinds of health interventions that are currently gatekept by price. When he describes the 2070 vision - where a peasant today lives better than a king fifty years ago - health is a big part of that calculation.

The Career as a Portfolio

There's a specific way Peechu thinks about his own career that's worth sitting with. A 40-year run in venture capital, in his framing, is about 10 real bets. Not fund cycles. Not individual investments. Ten moments where you commit fully to something uncertain, where the outcome will define whether you compounded or just kept moving. Most of those bets might not work. The ones that do have to work massively - they have to retain value for years, produce genuine outliers, and create something that stacks on itself.

That's a very different way of thinking about a career than "build a track record." It's closer to how he thinks about the companies in his portfolio. Go long. Be patient. Be present. Know when you're in year nine and when year twenty is what actually matters.

At the start of 2025, he's running Felicis's $900 million fund - a number that would have seemed impossible from the perspective of the $41 million Fund I in 2010. The portfolio spans fintech, health, AI infrastructure, enterprise software, defense tech, and developer tools. His Twitter handle is @speechu, which is both a pun on his last name and a kind of mission statement for someone who has spent fifteen years saying things clearly, publicly, to people who need to hear them.