Cursor's $60B sale sets the VC record; the fund-making check in it was $750,000

About 1,333 times money on the opening check, if the roughly $1 billion figure holds ^[11][3]. That is the number that gets screenshotted, and it is the wrong one to underwrite from. Two more checks followed the first, and Fortune's account puts no figure on either ^[2][10], so the blended multiple across everything BoxGroup put into Cursor is lower, and nobody outside the firm can say by how much.

The second ratio is more useful. A $1 billion outcome against a $60 billion purchase price implies the firm held something under 2% of Cursor at the end ^[12], several years after a 2022 seed bet ^[4]. For money that was among the earliest in ^[4], under two points is what the model costs: no board seat, and no fight to defend a percentage through later rounds ^[7]. The return came from the entry price and from staying on the line, not from the size of the slice.

That is why the constraint Jack Altman describes matters more than the anecdote does. There are only 100 points on a cap table, the Benchmark partner told Fortune, and as funds scaled up, the competition for those points pushed out the collaborative posture BoxGroup kept ^[8]. Breadth plus collaboration is a sourcing mechanism rather than a philosophy: back a lot of companies, never be the investor blocking someone else's allocation, and take deal flow back in return [c8b]. It works only while the larger funds tolerate a small early holder sitting there. The seed check that returns a billion is downstream of a round that let it in cheaply.

The underwriting detail worth keeping is that the product was wrong at entry. Truell pitched AI for computer-aided design, and Tisch says the decision was never about AI for CAD and always about the people ^[6]. What got paid, then, was a read on a founder, on a deal that a then-principal at the firm brought in ^[5]. In a business that has spent 15 years buying relevance through fund size and board seats ^[9], the winning input was a junior investor's conviction plus a check small enough that being wrong would never have been written up anywhere.

The limit is worth stating plainly. This is one outcome, and the figure that makes it legendary came from an unnamed source rather than a disclosure ^[3]. Firms have raised billions on thinner evidence than that. The real test is what Tisch does with a result that would comfortably support the multi-billion fund he says he does not want ^[7].