PitchBook's record $515.8B venture tally rests on OpenAI and Anthropic rounds

The team that renews its Miro contract next year will be renewing a product Bending Spoons expects to finish buying in the fourth quarter for $1.4 billion ^[15]. Miro raised its Series C at $17.5 billion, so the sale price is 8% of that valuation ^[27]. Airtable customers went through it first: Bending Spoons paid $1.3 billion for Airtable, down from an $11.7 billion valuation, or about 11% of the old figure ^[14][26].

The two AI labs' first-half rounds make up at least 38% of the nine-month record ^[25]. Third-quarter deal value fell about 40%, to $98.4 billion, with most of the drop coming out of venture-growth rounds ^[4]. Startups still closed an estimated 5,012 deals in the quarter, and PitchBook's records show only one busier quarter, at the start of 2022 ^[5]. The largest check was Databricks' $5 billion, which the report called "a far cry" from what the frontier labs raised earlier in the year ^[6].

PitchBook's bigger worry is getting the money back out ^[24]. Nizar Tarhuni, PitchBook's executive vice president of research and market intelligence, said "the real story sits on the exit side" ^[7]. In his view the IPO pipeline keeps slipping further out, leaving sellers to lean on mergers and acquisitions "to get anything done at all" ^[8]. SpaceX's $60 billion all-stock purchase of Anysphere, the developer of Cursor, made up 53.1% of third-quarter exit value on its own ^[9]. That puts the quarter's total near $113 billion ^[28]. Without that one deal, exits came to $53 billion, the lowest since late 2024 ^[10].

For a team holding contracts with venture-backed vendors, the practical question about each one is now who would buy it, and at what price. I'd sort a renewal list on two axes. The first is when the vendor last set a price. The report puts the median discount at 59% on Forge Global's secondary marketplace for shares of companies whose most recent funding round came in 2021 ^[16]. The second is how long your team would need to move off the product, counted in weeks of migration work.

A 2021-era price with a slow exit is the box that needs work before signing. The work is a data export someone has actually tested, and a term no longer than you would accept from an owner you have not met. That caution costs whatever discount a longer commitment would have bought. A recent price with a fast exit needs nothing beyond a current export. The other two boxes call for watching one thing each: a sale at the old-price, fast-exit vendor, and the next round at the new-price, slow-exit one.

Money for a team's own next round is concentrated too. Andreessen Horowitz alone closed funds worth $23.8 billion this year, roughly what the other 94% of new US funds, those under $500 million, raised between them ^[18][29]. The National Venture Capital Association's president and chief executive, Bobby Franklin, said that strong AI innovation "can obscure growing challenges within the fundraising market" ^[20].