Benchmark Capital, the storied Silicon Valley VC firm known for its early investments in eBay, Snap, Uber, and Twitter, is breaking with one of its signature traditions of keeping its funding at about $425 million and backing only young startups. After more than two decades of restricting vehicles to below that amount, the organization has completed $2 billion in commitments in two new funds, including a $1.25 billion vehicle dedicated to late-stage investments, according to the Wall Street Journal.
While the fund sizes of many venture capital firms have ballooned into the billions over the past decade, Benchmark has stuck to the strategy that helped make it legendary. The firm maintained a model designed to maximize significant returns for its limited partners by being selective about every startup it supported, typically taking a 20% stake.
However, Benchmark's relatively small fund size likely prevented the company from investing in capital-intensive AI startups, especially basic model makers, where round sizes often reach hundreds of millions. As a result, the company has not invested in Anthropic, OpenAI, or other capital-intensive AI labs such as Periodic Labs, Reflection AI, and Recursive SuperIntelligence.
Benchmark bet on AI with mixed results. The company led a $75 million round in Manus, a Singapore-based AI agent platform that reached $100 million in annual recurring revenue within eight months of launch. Late last year, when Meta agreed to acquire Manas for about $2 billion, it looked like a new benchmark winner would be born. But Chinese regulators blocked the deal in April, claiming the company, which was founded in China before moving to Singapore, violated export control laws, leaving Benchmark's shares in limbo.
Benchmark's new $750 million early-stage fund gives the company flexibility to write checks in an environment where early-stage valuations have skyrocketed. The firm has traditionally backed companies at the Series A stage, but has recently gained more flexibility to invest in companies at other early stages of development.
In recent months, Benchmark has backed two Series B startups. Gumloop, a platform that allows businesses to create AI agents without writing code, and Monaco, an AI-native sales and CRM platform.
Everett Rundle, general partner at Benchmark, previously told TechCrunch that the company is looking to “build meaningful, deep relationships with entrepreneurs, and that can happen relatively early in the company's lifecycle, at the seed stage.” [Series] Ah, so [Series] B. ”
As TechCrunch previously reported, the company stepped into late-stage investing when it raised a $225 million special purpose vehicle (SPV) to participate in Cerebras' $1 billion pre-IPO round. Benchmark first led the chipmaker's Series A in 2016. Cerebras held its IPO last month, and Benchmark returned $3.25 billion at the IPO price.
This windfall led the company to raise dedicated growth capital. The new vehicle will involve five to six large investments in both existing portfolio companies and startups, according to people familiar with Benchmark's strategy.
The two new funds aren't the only changes at Benchmark. Over the past two years, the company has experienced significant changes in general partners.
In 2024, Miles Grimshaw leaves the company and returns to Thrive Capital. And last year, Sarah Tavel, Benchmark's first and only female general partner to date, took on a less involved venture partner role, while Victor Lazarte left to start his own venture capital firm.
Benchmark, which traditionally operates with four to six general partners, has added two new high-profile investors to its team to fill its ranks. Randle, who was hired from Kleiner Perkins, and Jack Altman, the younger brother of OpenAI CEO Sam Altman. The move suggests that even Benchmark, long defined by its resistance to growth, believes the AI era requires a different strategy: more capital, more stage, and fresh blood at the partner table.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.