Cerebras and Fervo did not just price in the week of May 11 through May 17. They reset the ceiling for what the 2026 IPO market will fund. Cerebras used a May 11 S-1/A to lift its range to $150 to $160 and expand the deal to 30 million shares, then priced at $185 on May 13. Reuters said the stock opened 89% above issue on May 14. Fervo followed with its own May 11 reset, raising the range to $25 to $26 and upsizing to 70 million shares from 55.6 million, then priced at $27 for $1.89 billion. Reuters said the shares opened at $36 and valued the company at $10.21 billion in the Nasdaq debut.
That pair matters because both deals repriced higher after amended filings, not because the market blindly embraced anything with an IPO ticker. IPOGrid reads the week as a reopening at the top end of quality and narrative strength: AI infrastructure and power for AI found eager buyers, while the rest of the calendar still had to clear a much harder bar on leverage, structure, and sponsorship.
Large Deals Carried the Tape
Cerebras raised $5.55 billion at pricing, which made it the week’s dominant event on size alone, but the more important signal was the last-minute price discovery. A company that had already moved its range from $115 to $125 up to $150 to $160 still priced another step above the revised range. Our interpretation is that this was not merely a hot AI tape; it was evidence that investors were willing to pay up for one of the few scaled, public-ready AI infrastructure names with real revenue. The reviewer’s concern is not demand but aftermarket psychology: an 89% opening pop is the kind of move that validates the deal and complicates follow-on discipline at the same time.
Fervo was the cleaner infrastructure read-through. Its May 11 amendment lifted both price and size, and the company’s final terms came in above the marketed range. Reuters tied the first-day move directly to data-center and AI-linked power demand, which is exactly why the deal deserves attention beyond the renewable-energy bucket. Fervo offered the market a rare combination of growth, policy relevance and an energy-security narrative that does not depend on oil. IPOGrid would frame that as a sign that the market is willing to stretch for infrastructure stories when the demand thesis is legible in one sentence.
The Selective Middle
Below the headline deals, the tape got more judgmental. GMR Solutions cut its IPO price from an earlier $22 to $25 expectation down to $15 in its May 12 S-1/A and simultaneously disclosed a $500 million concurrent private placement alongside the $478.7 million IPO. Reuters reported that the stock opened at $13.50, down 10% from the offer price. That is a useful counterpoint to Cerebras and Fervo: the market will still fund a deleveraging story with real scale, but only after a severe valuation reset. Our read is that investors treated GMR as balance-sheet repair first and growth equity second.
EagleRock held a more conventional $17 to $20 range, priced 17.3 million shares at $18.50, and raised about $320 million. The follow-through looked better too: Reuters described a positive NYSE debut that pushed the company toward a $3 billion valuation. We would separate that from a generic energy trade. EagleRock sold a royalty-and-access model tied to the Permian rather than an exploration balance sheet, and that cleaner structure appears to have mattered in a week when investors were willing to own energy exposure but not every energy wrapper.
The small end of the calendar remained much less forgiving. Exyn marketed 2.5 million units at $7.75 to $8.75, each unit carrying one share and one warrant, then priced at the bottom at $7.75 for roughly $19.4 million. Micware marketed 2.15 million ADSs at $7 to $9 and then closed an upsized 2.85 million ADS deal at $8 for $22.8 million. Those are not interchangeable outcomes, but they fit the same pattern: smaller issuers still needed warrants, ADR packaging, or both to get done. The reviewer’s concern is not that these deals priced; it is that they did so in forms that still advertise fragility.
What Still Looks Unsettled
The week’s effective and priced fringe names were also a reminder that paper can move forward without becoming investable in the same way. ReserveOne’s May 13 424B3 and effectiveness notice pushed the crypto-reserve merger structure further along, but our interpretation is that the structure remains the story, not operating fundamentals. Phoenix Energy One’s May 14 424B3 did something similar for a self-underwritten preferred-share offering whose debt-repayment angle is easier to understand than the aftermarket setup.
On the watchlist, Lincoln International’s roadshow launch for 21,049,988 shares at $18 to $20 was the cleaner traditional filing signal. If the market is looking for a next-wave test case after the week’s blockbusters, that kind of sponsor-backed but otherwise standard underwriting setup may tell more than another micro-cap unit deal.
The bottom line is that the window widened, but it did not democratize. Cerebras and Fervo showed that investors will chase scarcity and thematic clarity. GMR showed that they will still punish leverage-heavy valuation asks. EagleRock showed that a simpler asset story can work. Exyn, Micware, ReserveOne and Phoenix showed that the lower end of the market is still relying on structure to compensate for weaker natural demand. That is a better week for issuance than the headline count alone suggests, but it is still a selective market, not a fully open one.