Benchmark breaks its own rule with a $2bn raise and a first growth fund

Benchmark has closed $2bn across two new funds, including a $1.25bn growth fund, its first dedicated to later-stage investments, alongside a $750m early-stage vehicle. The move breaks the firm's long-standing discipline of keeping funds around $425m and investing only in young companies, a shift driven by the capital demands of the AI era. The new funds follow a $3.25bn return from AI chipmaker Cerebras' IPO, which prompted the growth vehicle, and signal Benchmark's adaptation to a market where smaller fund sizes had become a competitive constraint.

The discipline was the whole identity. For more than twenty years Benchmark kept its funds around $425m and backed only young companies, taking a roughly 20% stake in each and trusting that selectivity, not scale, would produce the returns. That model has now bent. The firm has closed $2bn across two new funds, according to the Wall Street Journal, including a $1.25bn vehicle dedicated to later-stage investments, the first growth fund in its history. Alongside the $1.25bn growth fund sits a $750m early-stage vehicle, larger than the funds Benchmark has historically raised and a concession that even seed and Series A cheques now have to be bigger to compete. The firm that defined itself by staying small has decided that staying small had become a constraint. It is not hard to see what the constraint cost. Benchmark’s fund sizes likely kept it out of the capital-intensive AI labs, where rounds run into the hundreds of millions; the firm never invested in OpenAI, Anthropic, or the wave of foundation-model start-ups that have defined the cycle. Where it did place AI bets, results were mixed. Benchmark led a $75m round in Manus, a Singapore-based agent platform that reached $100m in annual recurring revenue within eight months, and watched Meta agree to buy it for roughly $2bn, only for Chinese regulators to block the deal in April over export-control law, leaving the stake in limbo. The new flexibility shows in recent deals. Benchmark has traditionally entered at Series A, but in recent months backed two Series B companies, Gumloop, a no-code platform for building enterprise AI agents that raised $50m from the firm, and Monaco, an AI-native sales and CRM platform. General partner Everett Randle has framed the shift as a matter of relationship rather than stage, telling TechCrunch the firm wants a deep relationship with founders that can begin at seed, Series A, or Series B. The growth fund has a clearer origin: a windfall. Benchmark first led Cerebras’s Series A in 2016 and later raised a $225m special-purpose vehicle to join a $1bn pre-IPO round. Cerebras went public last month https://thenextweb.com/news/cerebras-ipo-5-55-billion-biggest-tech-2026 and returned the firm $3.25bn at the IPO price, and that result, by the account of a person familiar with the strategy, prompted the dedicated growth vehicle. It will make five or six large investments across existing portfolio companies and new ones. The funds are not the only thing that has changed. Benchmark has reshaped its partnership over two years, with Miles Grimshaw leaving for Thrive Capital in 2024, Sarah Tavel moving to a venture-partner role, and Victor Lazarte departing to start his own firm. In their place the firm added Everett Randle from Kleiner Perkins and Jack Altman, brother of OpenAI’s Sam Altman. Put together, the changes describe a firm rebuilding for a different game. More capital, more stages, new partners. Benchmark spent two decades arguing that restraint was the edge. The $2bn it has just raised is an admission that, in the AI era, restraint had started to look like a handicap. Get the TNW newsletter Get the most important tech news in your inbox each week.