Analysis
I keep hearing that the IPO window is open. It is open the way a nightclub is open -- there is a line, and a list, and you are not on it.
The aggregate data supports the optimists. U.S. IPOs raised $34.2 billion through May 31, up 163.9% year over year. OpenAI filed confidentially with the SEC in June for what would be one of the largest debuts in market history. Anthropic has been working with banks on its own filing. Reddit joins the S&P 500 before the open on August 18, the second pure-play social company to make the index after Meta. Figma grew revenue 48% to $370 million in Q2. That is a functioning market.
Now look at the distribution. Reddit was down 31.2% year to date before the index news, near $175 a share and roughly $29.5 billion in market cap, and it took an S&P inclusion event to move the stock 15%. That is a technical bid, not fundamentals being rewarded. Meanwhile the filings actually crossing the SEC's desk most weeks are things like BayFirst Financial's S-1 and a blank-check vehicle.
“IPOs raised $34.2 billion through May 31, up 163.9% year over year.”
My read is that we have a two-tier market and we are describing it with one number. Tier one is four or five companies -- OpenAI, Anthropic, SpaceX, Stripe, maybe Databricks -- where the listing is an event and demand is guaranteed. Tier two is the several hundred venture-backed companies with $50 million to $200 million in revenue that were supposed to go public in 2022, did not, raised a bridge, and are now being told to grow into a mark set four years ago. For them the window has not opened at all. It has been replaced by a secondary market and an acquisition path -- which is exactly why Stripe just paid over $7 billion for OpenRouter rather than watching it list.
The practical consequence for GPs is that DPI does not arrive through the tier-one names unless you were on those cap tables. It arrives through trade sales at prices set by strategic buyers with their own agendas. Plan the fund accordingly.
Room for disagreement: the strongest counter-argument is that this is simply what every reopening looks like. The 2013 and 2020 cycles also started with a handful of marquee names and broadened over four to six quarters as bankers rebuilt confidence and comparable trading data accumulated. If OpenAI lists well, the tier-two backlog gets repriced upward almost mechanically, and a year from now the concentration I am describing will read as a starting gun rather than a ceiling. I think the difference this time is the sheer size of the tier-one names relative to everything else -- but that is a judgment, not a fact, and the next four quarters will settle it.
One more number, because it is the one that changed my mind. Reddit's 15% jump on index inclusion, reported by Forbes, came on flows rather than earnings. When the marginal buyer of a profitable, growing internet company is an index fund rebalancing rather than an active manager taking a view, the price discovery that a healthy IPO market depends on is not happening.