Analysis
When Goldman Sachs forecast in February that US IPO proceeds would quadruple to a record $160 billion in 2026 across roughly 120 listings, it read as an aggressive, headline-grabbing call built on a recovering economy and a reopening tech IPO window. Five months later, with the biggest names on the calendar still to come, that number is starting to look conservative rather than bold.
The math is not subtle. SpaceX's IPO alone raised $75 billion -- the largest venture-backed listing in history -- and that single transaction accounts for nearly half of Goldman's entire full-year target. Add SK Hynix's $26.5 billion US listing earlier in July, the largest foreign IPO in US history, and CXMT's $8.6 billion Shanghai debut, and just three transactions this year already approach $110 billion in combined proceeds, before counting the hundreds of smaller listings that make up the rest of the 208 US IPOs priced through July 27 -- itself already running about 6% ahead of last year's pace at the same point in the calendar.
This week alone illustrates how concentrated and relentless the pace has become: Zhongji Innolight's roughly $8 billion Hong Kong offering and Jersey Mike's up to $1.09 billion US listing both landed in the same seven-day window, alongside Scribe Therapeutics' $128.7 million biotech debut. That's not an unusually busy week by 2026 standards -- it's closer to the new normal, with multiple billion-dollar-plus offerings pricing simultaneously on a near-weekly basis since the second quarter.
“GPs should be modeling exit environments off the median deal size, not the headline aggregate, when advising portfolio companies on IPO timing.”
What Goldman's forecast likely underestimated in February is precisely the pattern this issue has documented all year: a small number of enormous, AI-adjacent companies -- chipmakers, AI labs, data center and networking suppliers -- going public at valuations that dwarf traditional IPO math, while the broader base of smaller listings simply continues at its normal pace underneath them. The $510 billion in H1 2026 global VC funding this outlet covered recently showed the same concentration dynamic in private markets, with OpenAI and Anthropic alone capturing 43% of the total; the public markets in 2026 are exhibiting the identical pattern, just with SpaceX, CXMT and Zhongji Innolight playing the concentrating role instead.
For founders eyeing a 2026 or 2027 listing, the practical lesson is that the "average" IPO environment implied by Goldman's $160 billion full-year number bears very little resemblance to what any individual company outside the AI-infrastructure megadeal tier should expect -- the real median listing this year looks much closer to Scribe Therapeutics' $128.7 million than to SpaceX's $75 billion. GPs should be modeling exit environments off the median deal size, not the headline aggregate, when advising portfolio companies on IPO timing.
What to watch: whether Goldman or other banks revise their 2026 full-year IPO forecast upward given the pace through July, whether the concentration in a handful of megadeals continues to intensify or broadens out to more mid-sized listings in the second half, and whether Anthropic, OpenAI or Databricks add another SpaceX-scale data point to the year's total before December.
