The IPO window is reopening—but market plumbing will decide whether it lasts

The U.S. equity-issuance machine is moving from scarcity toward supply. That is constructive for companies that need public capital, but it also creates a more demanding test for investors: can new listings, follow-ons, lockup releases and buybacks coexist without overwhelming liquidity or exposing weak demand?

The thesis: reopening, not a free pass

The first half of 2026 produced an unusually strong equity-capital-markets backdrop. Houlihan Lokey reported $297.1 billion of U.S.-focused equity-capital-markets proceeds in the first half, with IPO deal count nearly doubling year over year.^[1] EY’s review provides the more useful qualification: 62 U.S. IPOs had raised more than $50 million through June 30, versus 34 in the comparable 2025 period, while 12 companies raised more than $1 billion, compared with four a year earlier.^[2]

That is a meaningful change in supply, but not yet proof of a durable, broad-based cycle. EY says the rebound has been concentrated in AI and AI-adjacent issuance, with aerospace, defense and biotech also active; enterprise software has remained relatively muted as investors assess how AI changes software business models.^[2] The base case is therefore conditional: the window can stay open if earnings quality, pricing discipline and sector breadth keep improving. It can close quickly if rates, geopolitics or AI-spending doubts reset risk appetite.

Why issuance and liquidity matter now

A public market is not just a venue for the first sale of shares. It is a continuing system for absorbing supply and demand:

The practical distinction is between capital formation and liquidity absorption. A large deal can raise substantial money and still trade poorly if the aftermarket float is thin, the valuation is difficult to underwrite, or early holders sell into limited demand. Conversely, a steady calendar of smaller, well-priced offerings can improve price discovery by giving investors more opportunities to compare business quality.

The fall calendar is the next test

Renaissance Capital’s September 8 fall preview described a pipeline in which AI companies are expected to take center stage, after record year-to-date proceeds.^[3] Its week-ahead report also pointed to up to eight companies that could begin roadshows after Labor Day, while noting that only one small direct listing was scheduled at that point.^[3] The difference between a roadshow pipeline and a priced deal is important: filing, marketing and pricing are separate steps, and a company can wait if the market deteriorates.

That optionality is a feature of the current market. Companies with strong cash positions do not have to accept weak pricing, while sponsors and existing shareholders may prefer a follow-on, secondary sale or delayed lockup release over a rushed primary offering. The result can be a calendar that looks full but remains highly sensitive to the tone of the tape.

The plumbing is changing beneath the calendar

The Securities and Exchange Commission approved a temporary amendment to the national-market-system volatility plan establishing price-band protections for overnight trading.^[4] This is a market-structure development, not a forecast: it changes how an extreme move may be handled, but it does not tell us whether demand for a new issue is healthy.

Exchange incentives matter too. A September 1 Federal Register notice described Nasdaq’s proposed changes to exchange-traded-product listing fees and a new quality-liquidity-provider program.^[4] Separately, NYSE rule filings in 2026 included changes involving designated-market-maker economics for IPOs and transfers.^[4] These details are easy to overlook, but they shape who is willing to quote, when liquidity appears and how efficiently investors can trade around new supply.

The implication is straightforward: issuance statistics should be read alongside spreads, depth, turnover, price gaps and post-IPO performance—not in isolation.

Testing the growth-and-demand hypothesis

The supplied hypothesis spans cloud software, consumer names, crypto, AI infrastructure, industrial electrification and smaller speculative companies. The current evidence supports parts of it, but it does not validate the whole basket equally.

The strongest operating-demand evidence in this pass comes from NVIDIA’s recent earnings-call record. In its August 26, 2026 call, management described demand for training and inference compute as exceptionally strong, while also saying frontier AI companies’ growth was increasingly constrained by access to compute and financing capacity.^[5] In the same research record, management said sovereign AI, neoclouds, startups and enterprise demand were broadening beyond the largest cloud customers.^[5]

That supports the demand side of the hypothesis for AI-infrastructure-linked names such as NVDA, AMD, GEV and ETN, but it also raises a financing question: demand can be real while the capital intensity needed to serve it becomes a source of risk. EY similarly identifies AI investment as both a major opportunity and a potential source of investor concern.^[2]

The market is already differentiating rather than moving as one block. In pre-market trading on September 9, DDOG was $210.90, up 0.32% versus its September 8 regular-session close as of 8:49 ET; SNOW was $335.45, essentially flat at -0.01% as of 8:51 ET; COIN was $182.40, up 1.93% as of 8:52 ET; and AMD was $500.54, down 1.03% as of 8:52 ET.^[6] Those are snapshots, not conclusions, but they illustrate why a broad “resilient demand” label is insufficient: each company still faces its own earnings, valuation, financing and liquidity test.

Upcoming reports will provide scheduled checkpoints. RH is listed with an estimated September 10, 2026 after-close report; COIN with an estimated October 29 after-close report; and NVDA with an estimated November 18 after-close report.^[7] The dates are calendar estimates, not confirmed company announcements, and should be treated accordingly.

What would confirm or weaken the reopening?

Evidence that would confirm the constructive case:

• More IPOs pricing across software, industrials, healthcare and consumer sectors—not only AI-adjacent companies.
• Stable or improving first-week and first-month aftermarket trading without relying on unusually small floats.
• Follow-on and secondary deals clearing with transparent discounts rather than abrupt price gaps.
• Lockup releases being absorbed without persistent deterioration in volume-weighted liquidity.
• Buybacks remaining a source of corporate demand while companies preserve balance-sheet flexibility.
• Earnings from the hypothesis set showing durable demand rather than only capacity expansion and capital spending.

Evidence that would weaken it:

• A return to a handful of mega-deals carrying most proceeds.
• New issues pricing only after substantial discounts or shrinking deal sizes.
• Wider spreads, thinner depth or repeated volatility interruptions around listings.
• AI infrastructure demand remaining strong in headline terms while customer financing, power and deployment constraints rise.
• Consumer and software earnings failing to support the resilient-demand narrative outside the infrastructure complex.

What to watch next

1. The post-Labor-Day pricing tape: distinguish companies that merely enter the roadshow from those that price, trade and build a stable shareholder base.
2. The mix of capital raised: separate primary proceeds used by companies from secondary sales that primarily provide liquidity to existing holders.
3. Lockup supply versus public float: monitor released shares as a percentage of tradable float, not just shares outstanding.
4. Market depth around volatile sessions: the new overnight price-band framework may limit some prints, but depth and replenishment will reveal whether liquidity is genuinely improving.
5. Earnings quality across the hypothesis set: look for recurring revenue, conversion to cash, customer concentration, capex intensity and evidence that demand survives tighter financing conditions.
6. Breadth beyond AI: a durable IPO cycle should eventually include more ordinary businesses with predictable earnings, not just the market’s most compelling growth narratives.

The reopening is real enough to measure, but too concentrated to take for granted. The next phase will be decided less by the number of companies filing than by whether the market can absorb supply, preserve two-sided liquidity and reward durable earnings across more than one theme.

This article is research and education, not personalized investment advice.