Benchmark has raised $2 billion across two funds, but its five partners say they aren’t picking up their investing pace. Each still expects to make just two or three new bets a year. The bigger change is where the firm can invest: alongside a larger core fund, it now has a $1.25 billion growth fund for opportunities its early-stage mandate previously put out of reach.
At this week’s TechCrunch Disrupt, the partners said they want more investment capacity without taking on the machinery of a larger venture operation. Monday meetings still have no agenda. There’s no formal rule on whether one partner can veto a deal. Despite economic warning signs, the group seemed relatively untroubled, viewing many of those signals as products of hysteria fueled by X.
The people have changed more than the operating model. Three of the four partners who took part in TechCrunch’s interview three years ago are no longer at the firm. This summer’s fundraising also marked only the second departure in Benchmark’s 31-year history from raising a much smaller vehicle.
Can that small-team discipline carry over into growth investing? Peter Fenton’s answer starts with concentration, not more deals.
Asked how many investments the growth fund might make, Fenton said the partners had different answers. His preference was “four to six, maybe seven to eight.” Citing the J.P. Morgan paper on the agony and ecstasy of concentrated investing, he said a successful fund would likely get 90% of its returns from one or two investments.
For Fenton, the core business remains working closely with founders from the beginning, usually for 10 to 15 years. That shoulder-to-shoulder commitment doesn’t scale, he said. The core fund had stayed at $500 million for 20 years; increasing it now reflects inflation and the size of the economy. He said not making that adjustment earlier was probably a tactical error.
The growth fund represents a more substantial expansion of the mandate. Fenton said the ecosystem had given Benchmark permission to move beyond early-stage investing, and the old constraint had cost it opportunities. Some investments that came its way, he said, could have returned between 10 and 1,000 times its money but didn’t fit its early-stage focus.
A bigger fund doesn’t mean a bigger decision-making machine. Ev Randall, whose previous firms include Founders Fund, Bond, and Kleiner Perkins, explained the difference in organizational terms. Some venture firms have 50- or 60-person investment teams, hierarchies, and investment processes divided into product groups. Small partnerships allow more direct, higher-fidelity conversations without decisions passing through a Byzantine corporate structure. He compared it to a sports team taking the field together.
Randall said he had been drawn to that model early in his career. But size alone wasn’t the attraction at Benchmark: the current partners didn’t merely like one another, they admired and learned from one another. He called that combination rare and said it made him want to join immediately. He believed fellow newcomer Jack Altman felt similarly.
Altman joined in February, despite having raised a second, $275 million fund for his own firm just a year earlier. He had told The Wall Street Journal that the raise took one week. With a hugely popular podcast and an established solo investing operation, he wasn’t joining because he lacked an independent platform.
He traced the decision to recording a podcast with the Benchmark partners at the start of this year. Their commitment to a small fund mattered, but what stood out more was their desire to be the first people founders called, stay concentrated, and help companies. In an industry often caught up in money and status, he felt the partners genuinely cared about founders. That impression grew tenfold as he got to know them.
Investing individually had been enjoyable, Altman said. In a market this unsettled, though, he wasn’t sure how he would feel trying to work everything out alone. Having smart, focused colleagues to learn with and from was a major part of the appeal.
The partnership’s Monday meetings offer a concrete example. Chethan Puttagunta joked that Altman had tried to introduce an agenda. Altman replied: “I’ve got an agenda for next Monday.”
Eric Vishria described the meetings as one ongoing conversation rather than a scheduled sequence of decisions. The partners discuss companies they’re meeting, what caught their attention and why, and what they’re hearing in board meetings. During rapid change, the aim is to work out what has become possible that wasn’t possible six months earlier, and which founders are making it happen.
Five people who enjoy spending time together don’t need to systematize that exchange, Vishria said. He described it as a decades-long conversation. After the Disrupt session, the partners were due to have dinner with an entrepreneur who had built a company worth multiple hundreds of billions of dollars. Benchmark wasn’t an investor; the purpose was to stretch their thinking.
Leaving space also lets the team respond to portfolio companies, Puttagunta said. If a founder needs help recruiting an executive or securing a strategic partnership, all five partners can make themselves available.
Even deal approval is less formal than outsiders might expect. Asked whether one person could block an investment in the equal partnership, Puttagunta said: “Nobody knows. We don’t have rules around it.” Altman said he didn’t know either and had been waiting to see what someone else did.
Altman argued that a low volume of deals makes this lack of structure useful. A firm investing in 100 or 200 companies annually would need a customer relationship management system and a formal voting process. Benchmark’s infrequent investments don’t require the same machinery.
Fenton added that the firm doesn’t approach pitches as an exercise in confidently labeling someone a winner or loser. “It’s not Shark Tank,” he said. Instead, he described a sense of awe at the possibilities. Still, an entrepreneur can feel when someone in the room dislikes them; in those polarizing cases, he said, Benchmark wouldn’t make a great partner.
Nor does faster company growth mean Benchmark intends to increase its deal count. Puttagunta said the pace remained consistent and accepted that the firm wouldn’t participate in every successful company. Its job is to identify an entrepreneur with a particular strength and find a way to amplify it. How much capital that person wants, the investment structure, or where they are becomes something to work through, rather than the starting point.
Founder commitments face a test in a faster-moving AI market. As AI companies expand into one another’s territory, what happens when backing a founder means backing a competitor to another portfolio company?
Vishria said Benchmark’s ethos rests on a commitment to an entrepreneur and close involvement in what they’re building. That commitment should matter to both sides. But companies now evolve, pivot, and change direction quickly enough that conflicts will occur, even when none existed at the original investment.
Maintaining confidentiality and exclusivity within an area remained important, he said. Benchmark had been fortunate not to encounter the kind of extreme direct conflict illustrated by the recent allegations involving rival companies Factory and Cognition. His position wasn’t that the old commitment had disappeared, but that the landscape around it was changing.
Puttagunta expects more collisions. In his view, founders are especially ambitious in the AI era, and as the cost of coding new products effectively approaches zero, they see more territory they want to pursue. He argued that investors can still maintain relationships and behave respectfully when companies begin running into one another.
Altman’s first Benchmark investments illustrate how the partnership can also broaden an individual investor’s range. His background was in software, but since joining he has invested in chips, robotics, and infrastructure. He said he had completed a few deals, including recent investments that weren’t all disclosed.
There wasn’t a formal onboarding process. Recalling a question about how onboarding had gone, Altman said his response was: “It was perfect. There was no onboarding.” Either Puttagunta or Vishria, he couldn’t remember which, had told him not to get tense about his first investment. He would make many investments at the firm, venture returns follow extreme power laws, and the odds of his first deal being the defining one were low.
Altman said he trusted that the partners understood his character and investing style, and he tried to stay consistent with that. Their involvement also mattered in practical terms: he didn’t think Benchmark would have won his investments without his colleagues, and wasn’t sure he could have reached a yes without them.
Cerebras provides a longer-running example of that shared judgment, including its reversals. Vishria said Benchmark invested in the AI chip company’s first round in the spring of 2016, after a decade without making a hardware investment. The founders seemed good and had a simple thesis, but he didn’t understand much of the rest.
On the Sunday night before the founders returned to pitch the partnership, Fenton told Vishria that the investment went against everything he had learned in venture capital. After Monday’s pitch, Fenton changed his mind. Vishria said understanding one another’s instincts was a major part of the process.
That decision began a journey with many ups and downs, including several points when the company could have died, according to Vishria. He emphasized that the journey was far from over. Falling in love with a founder and an idea, he said, is why the partners do this work.
Fenton described Andrew and his co-founders as heretics with an unorthodox thesis. He connected that to criticism of Cerebras’s stock, arguing that polarizing companies are exactly where venture investors should be looking. “If it’s consensus, why is there a startup?” he asked.
The interview then turned to a concrete test of that conviction: Cerebras had priced its public offering at $185 earlier this year, climbed to $350, and returned to near $185. The next question put to the partners was why.
Benchmark’s Randall argues that AI’s market swings say more about the information environment than customer demand. In his view, predicting the macro picture precisely is “always a fool’s errand,” but the way investors interpret news has changed substantially since he started investing.
Asked whether dwindling free cash flow at many AI companies, expensive borrowing and Character.AI’s bankruptcy could signal a severe correction, Randall pushed back on the bankruptcy’s broader significance. He described Character.AI as the remnant of a business whose founder and team had been acqui-hired into Google a year and a half earlier, and which had already been sold for $2.3 billion. Treating its Chapter 11 filing as evidence of an AI chill, he argued, reflected an information environment around public tech stocks increasingly shaped by a “hyperactive X echo chamber.”
Randall sees a different market vulnerability: how much rests on estimates about two private companies. He attributed current trading levels for CoreWeave, Nebius and similar businesses to OpenAI’s and Anthropic’s annual recurring revenue figures. He expects both will probably go public within the next 12 months. A huge amount of capital expenditure and investment, he said, hinges on assessments of those two companies’ current and projected health.
That dependence doesn’t mean he sees weak underlying demand. Randall said Benchmark’s companies can’t get enough compute to serve customers willing to pay at good gross margins. He argued that nothing is inherently unhealthy about the AI ecosystem; the problem is volatility driven by data points that are often false or exaggerated to attract clicks. Taking OpenAI and Anthropic public would help, he suggested, by reducing the market’s tendency to swing 5% to 10% whenever an anonymous account revises an estimate.
Public markets bring their own demands, though. Vishria said that about six weeks before an IPO, he showed employees stock charts for CoreWeave, Datadog, DoorDash and Amazon to prepare them for volatility. He described how many stocks can move 500% between their highs and lows within a year, in either direction. His message was to focus on what the company can control and update public investors every 12 weeks. “Being a public company takes a stomach of titanium, and the employees have to understand that.”
The listing route matters, too. Former Benchmark colleague Bill Gurley has long argued that the traditional IPO process is broken, with the first-day price pop benefiting bankers and their clients rather than the company. His preferred alternative is a direct listing.
Vishria said he thought he might be the only person in the group who had worked on a direct listing, and agreed that the approach can make sense. But he considers it unlikely for the current crop of AI companies because most need substantial capital. IPOs raise money: Cerebras raised $6 billion, he noted. For companies with funding needs on that scale, the ability to bring in fresh cash weighs heavily.
OpenAI’s and Anthropic’s influence extends beyond market pricing to geography, with both helping pull the industry’s center of gravity back toward San Francisco. Asked whether founders now need to be there, Altman stopped short of calling it a requirement. He nevertheless described the city as, by far, the easiest place right now to find talent, capital and customers.
Probably everyone on the Benchmark team works with companies outside San Francisco, Altman said. But the partners themselves are there every day, giving the firm’s setup a strong geographic bias.
Fenton put numbers behind that preference, saying that historically more than 90% of capital returns have come from companies founded there. He acknowledged successful exceptions, specifically Shopify and Ramp. His explanation was the ecosystem effect: successful companies produce experienced executives and ambitious people, who then raise the bar for the next businesses. He doesn’t believe that process has happened systematically anywhere else quite as it has in San Francisco.
Randall’s recent meetings point in the same direction. He struggled to recall a non-San Francisco company he had met in the past 18 months that wasn’t actively spending time there. Those companies were opening an office, spending half their time in the city or planning to move there. His case for proximity came down to peers pushing one another: “Iron sharpens iron.”
