Aydin Senkut on Felicis Ventures: early Google, 53 unicorns, and why generalists beat experts in VC

Summary

Aydin Senkut founded Felicis Ventures after being turned down by five firms that told him he wouldn't even make a passable VC. He wasn't an engineer, wasn't senior enough at Google, and had no venture track record. Twenty years later, Felicis has backed 53 companies that exceeded $1 billion in valuation, logged 10 IPOs, and crossed roughly 100 exits — none of those unicorns, Senkut notes pointedly, were led by a domain expert.

His path into venture ran through Google, where he joined as the company's first product manager and first international sales manager. Working directly for Larry Page shaped his instinct for outliers: Page's standard response to any initiative was that it was roughly 1% of what he expected, done 10x too slowly. Senkut's attempt to manually launch Google in international languages famously frustrated Page enough that it accelerated the push toward machine translation — what Senkut calls his accidental contribution to Google Translate.

Early track record

His first fund was a solo GP vehicle of $4.5 million. The first institutional fund, raised in 2010, was $41 million. His first billion-dollar exit was Meraki, a Wi-Fi startup in Mountain View backed alongside two other ex-Googlers. His first IPO was Shopify, which went public at a $2.7 billion valuation — and has since grown roughly 50x from there. Google itself went 40x from its IPO price.

Recent bets

Felicis has participated in two rounds of Mercor, which went from zero to $100 million in gross revenue in 11 months by building an AI-driven platform for talent sourcing and expert interviewing. Senkut frames it as a new category of company — one that prior cycles simply couldn't have produced at that speed.

Superbase is the sharper underwriting story. Felicis invested at what was effectively a growth-round valuation when ARR was below $1 million, a price other VCs refused and several redirected into competitors at half the valuation. Since that investment, Superbase has grown 50 to 80x-plus in a few years. Senkut's read was simple: the most valuable tech companies have historically been database companies, there was a clear gap in developer-friendly backends, and the team was building on Postgres with early but real developer traction.

What's different about the AI cycle

Previous software waves had a ceiling — companies could capture maybe 10% of a customer's budget, and value creation still required humans to act on the software's output. AI closes that loop. Data entry, output, and action can all be automated, which means AI companies are going after labor budgets that can represent 40 to 70% of total company spend depending on sector. The addressable dollars are structurally larger than anything software-as-a-tool could reach.

The competitive structure question is real, though. Senkut acknowledges that high growth alone is insufficient — durable growth requires moats. His preferred moat is proprietary data: when a company holds data no one else can replicate, the commodity AI layer beneath it matters less. He points to a recent healthcare AI investment where the company unlocked a patient segment that had previously been unprofitable to serve; once hospitals saw that, they stopped splitting referrals across 19 vendors and directed 100% to the single company that could handle the hard cases.

Generalist vs. expert

The counterintuitive case against sector expertise runs through all 53 unicorns: deep experts anchor on how industries work today and are structurally skeptical of first-principles disruption. A generalist evaluating the same pitch asks whether the new approach works, not whether it matches convention. Senkut argues the same logic applies to founders — the best co-founding teams carry breadth across marketing, finance, product, and engineering rather than deep mastery in a single lane.

What he does value is exposure to outliers early in a career. Six years at Google, where nothing was done the standard way and the results were consistently non-linear, trained his pattern recognition for what genuinely exceptional execution looks like. That calibration, combined with a strong network and the discipline to invest consistently through down markets, is his framework for longevity in the business.

On cycle discipline: Felicis kept investing through 2021 and 2022 when peers paused. His argument is that venture returns are not determined by entry price precision — they are determined by whether a company can grow 50 to 100x. If it can, being modestly wrong on price barely registers. If it can't, even a bargain price doesn't save you.