Benchmark's $2 Billion Bet: Slow, Small and Staying in San Francisco

By Amara Osei

Benchmark raised $2 billion across two funds this summer — only the second time in its 31-year history it has broken from its far smaller vehicle. Yet at TechCrunch Disrupt this week, the firm's five general partners insisted they won't accelerate, even as rivals do.

The core discipline holds: two or three new bets per partner per year. Monday partner meetings still run without an agenda. And the partners dismissed many of the economy's warning signs as products of what Ev Randall called a "hyperactive X echo chamber."

Three of the four partners who sat for a joint interview three years ago have since left the firm. Jack Altman joined in February, a year after raising a $275 million fund for his own firm in one week, per the Wall Street Journal.

Why did the firm finally grow its funds?

Peter Fenton said the core fund's craft hasn't changed: working with founders from the beginning over a 10-to-15-year commitment, "shoulder to shoulder, and that doesn't scale." But he conceded Benchmark held the fund at $500 million for 20 years without adjusting for inflation — "probably a tactical error on our part."

The new $1.25 billion growth fund is an expansion of mandate. Fenton said missed opportunities "would have returned somewhere between 10 and 1,000x to Benchmark" under the old early-stage-only structure.

How concentrated will the growth fund be? Fenton wants four to eight investments. "If the fund is successful, one or two investments will generate 90% of the returns," he said.

What drew the newcomers?

Altman pointed to a podcast recording with the partners early this year. "So much in this industry gets tied up in money and status," he said. "What I felt from the Benchmark team then, and ten times more as I've gotten to know them, is it's a group that really cares about founders and helping companies."

Randall, who previously worked at Founders Fund, Bond and Kleiner Perkins, contrasted Benchmark's small partnership with hierarchical firms running 50- or 60-person investment teams. "You're going out on the field together, like a sports team," he said.

Can one partner kill a deal?

Apparently nobody knows. "Nobody knows. We don't have rules around it," Chethan Puttagunta said. Altman added: "I actually don't know. I was waiting to see what someone else did."

The low deal volume makes formal structure unnecessary, Altman argued. "If we invested in 100 or 200 companies a year, you'd want a CRM and a formal voting process," he said. "Because it's so rare, the lack of structure is a huge asset."

Eric Vishria said the Monday meetings work as "a decades-long conversation that only works if you're small and actually enjoy spending time together." After the Disrupt session, the partners planned dinner with one of the few entrepreneurs who has built a multi-hundred-billion-dollar company — someone they're not invested in — "just to stretch our thinking."

What about Cerebras and the AI selloff?

Vishria revealed Benchmark invested in Cerebras in spring 2016, its first hardware deal in 10 years, after Fenton flipped from skepticism overnight. The stock priced at $185 in this year's IPO, jumped to $350, and fell back near $185. Vishria showed the founders charts of CoreWeave, Datadog, DoorDash and Amazon before the listing: "Peak to trough, it's insane how many of them move 500% in a year."

He called direct listings unlikely for AI companies because they need capital. "Cerebras raised $6 billion. That matters."

On the first AI bankruptcy, Randall pushed back hard. "Character.AI was the leftover corpse of a company whose founder and team were acqui-hired into Google a year and a half ago, and it was already sold for $2.3 billion," he said. He argued CoreWeave and Nebius trading levels hinge on ARR figures of two private companies — likely OpenAI and Anthropic — that may go public within 12 months.

"There's nothing inherently unhealthy about the AI ecosystem," Randall said. "What's unhealthy is the volatility from data points that are often untrue or exaggerated for clicks."

Does a founder have to be in San Francisco?

Altman called the city "by far the easiest place to get talent, capital, and customers" — a strong bias, not a requirement. Fenton put a number on it: historically, more than 90% of Benchmark's capital returns have come from companies founded there, with Shopify and Ramp as the outliers.

Randall couldn't name a non-SF company he'd met in 18 months that wasn't opening an office, splitting time, or planning a move to the city. "Iron sharpens iron," he said.

The bet, in short: the frenzy changes everything around Benchmark, and the firm intends to change almost nothing.