GOVERNANCE WATCH
By InsidEntity Editorial Desk · Sep 26, 2026 · 13 min read
2026 is a year of very large listings, and none of them has a track record yet. This follow-up to The IPO Test runs its five questions on the year’s biggest names, and will be updated as deals price.
This is an InsidEntity research piece. It is not investment advice, and it takes no position on whether anyone should buy or subscribe to any listing.
2026 so far: record money, no verdict yet
US IPO proceeds are running at almost five times last year’s pace, on fewer deals. None of this year’s listings has a three-year record, so the 20 years of evidence in The IPO Test cannot grade them. What it can do is show which questions they raise.
The market in numbers. As of 26 September, Renaissance Capital counts 110 US IPOs priced in 2026, raising $146.9 billion. That is 29.9% fewer deals than at this point last year, but 393.7% more money. Counts depend on definitions: StockAnalysis, which includes smaller listings and SPACs, counts 245 as of 22 September. Neither uses the academic sample rules behind The IPO Test, so neither is directly comparable with its tables.
What the academic dataset already shows. Jay Ritter’s September 2026 edition records one IPO that doubled on its first day in the first quarter of 2026, Veradermics, against five in the whole of 2025. The companies choosing direct listings in 2026 have been microcaps. 2026 IPOs enter the dataset as a cohort only with its 2027 edition.
Is this 2021 again?
Not yet, but some of the same signals are back. The 2020 and 2021 listings are the most recent example of what happens when money, excitement and valuations peak together: big first-day gains, then three-year losses of 48.1% and 49.1%. The 2025 cohort, the newest with a full first-day record, shares some of those features and not others.
| 2020 | 2021 | 2025 | |
|---|---|---|---|
| US IPOs | 165 | 311 | 90 |
| Average first-day return | 41.6% | 32.1% | 29.3% |
| Closed below offer on day one | 19.4% | 24.4% | 24.4% |
| Gross proceeds | $61.9bn | $119.4bn | $39.0bn |
| Money left on the table | $29.7bn | $28.7bn | $13.1bn |
| Median public float | 22.0% | 16.7% | 14.5% |
Source: Ritter (14 September 2026), Tables 1, 1a, 19 and 21.
What matches. First-day gains are high again: 2025’s 29.3% was the fourth-highest of the past 11 years. Floats are thinner, with a median of 14.5% of shares sold to the public. Dual-class structures reached 41.1% of 2025 IPOs, above the 33.1% of 2021.
What doesn’t. The number of deals is far lower: 90 in 2025 against 311 in 2021, and 2026 is running 29.9% behind last year’s count. The surge in 2026 proceeds comes from a few very large listings, not from a rise in the number of companies coming to market.
Why it matters. Thin floats and dual-class shares leave outside investors with less weight in the company. The historical data does not show that dual-class IPOs returned less. It does show that the years with the biggest first-day gains were followed by the weakest three-year results. That is a reason for scrutiny, not a forecast.
The entry price, in 2026
Two of the year’s largest US listings show a point from The IPO Test in real time: the long-run evidence measures returns from the first-day close, not the offer price, and the difference can decide the result. Investors who received shares at the offer price are ahead. Investors who bought at the first-day close are behind.
These are share-price moves over three to four months, not three-year returns. They illustrate a mechanism; they are not evidence about how either company will do.
The listings under the spotlight
Each name is read against The IPO Test’s five questions: valuation, capital, financial quality, governance and growth burden. InsidEntity places each in a tier by how much scrutiny it needs now. Observation: already listed; the buying decision at the IPO is over. Watch: an offer is open or formally announced. Close Watch: the patterns the data warns about (a hot sector, record deal sizes, big first-day moves, rich valuations) are most concentrated here.
| Company | Sector | Where it lists | Tier | Status, 26 Sept 2026 |
|---|---|---|---|---|
| SpaceX | Space, satellite internet | Nasdaq | Observation | Listed 12 June at $135 |
| Cerebras | AI chips | Nasdaq | Close Watch | Listed 14 May at $185 |
| China Resources New Energy | Wind and solar power | Shenzhen | Observation | Listed early July at 10.11 yuan |
| Dangote Petroleum Refinery | Oil refining | Nigerian Exchange | Watch | Offer open to 13 October |
| Airtel Money | Mobile money | London | Watch | Pricing expected mid-October |
| OpenAI and Anthropic | AI models | Not announced | Close Watch | Confidential draft filings only |
The tier is a statement about scrutiny, not a prediction or a rating. The sections below run each name through the five questions. The historical figures give context; they do not forecast any of them.
SpaceX: the biggest listing, the biggest valuation burden
Tier: Observation. Listed on Nasdaq on 12 June at $135. On 22 September the shares closed at $151.85: 12.5% above the offer price, and 5.7% below the first-day close.
- Valuation. At roughly $1.75 trillion, the IPO valued SpaceX at about 94 times its 2025 revenue of $18.7 billion. The 46 large US IPOs priced above 40 times sales returned -44.8% over three years on average. No company of this size is in that sample, so this is a warning, not a forecast.
- Capital. The 555.6 million shares sold were new, according to the offering document, so the money went to the company rather than to selling shareholders.
- Financial quality. SpaceX listed as a loss-maker, with a 2025 net loss of $4.9 billion driven by AI and Starship spending. Starlink is the profitable segment, but its revenue per user is falling.
- Governance. InsidEntity’s CRR is 2.59, Caution. Elon Musk holds about 42% of the economic interest but more than 80% of the votes, and can be removed from his roles only by a class of shareholders he dominates. Auditor Independence scores 0.0 on tenure. See SpaceX’s full record on InsidEntity, or our earlier SpaceX analysis.
- Growth burden. At 94 times revenue, the price assumes years of rapid growth across launch, Starlink and AI. Supply is also rising: more than 700 million shares unlock in September and about 650 million in October, with Musk’s own shares locked for 366 days.
Cerebras: the profit that wasn’t cash
Tier: Close Watch. Priced 30 million shares at $185 on 13 May and closed its first day at $311. On 22 September the shares closed at $208.49: 12.7% above the offer price, and 33.0% below the first-day close.
- Valuation. At the offer price, Cerebras was valued at about $56.4 billion fully diluted, roughly 110 times 2025 revenue of $510 million. On the same fully diluted basis, the first-day close of $311 implied about $95 billion, roughly 186 times revenue (our arithmetic). Both are far above the 40-times line.
- Capital. All 30 million shares were new, sold by the company. The prospectus earmarks the money for general corporate purposes, including working capital, and about $329.6 million for tax-withholding obligations on employee share awards.
- Financial quality. This is where the prospectus repays reading. Cerebras reported 2025 net income of $237.8 million, but the figure included a large non-cash gain on a forward contract, and operating cash flow was negative $10.1 million. Two customers, MBZUAI and G42, accounted for about 86% of 2025 revenue.
- Governance. Class B shares carry 20 votes each. After the offering, Class B holders control about 99.2% of the voting power. Public shareholders have economic exposure but almost no say.
- Growth burden. Cerebras reported $24.6 billion of contracted future revenue, much of it tied to OpenAI, with about 15% expected within two years. The price assumes that backlog converts into cash, and that the customer base widens.
China Resources New Energy: a record pop on falling profits
Tier: Observation. Asia’s largest IPO of 2026 at the time it listed, it raised 24.5 billion yuan ($3.6 billion) on the Shenzhen Stock Exchange in early July. The retail tranche was 683 times oversubscribed, and the shares closed their first day up 137% at 23.95 yuan, against an offer price of 10.11 yuan.
- Valuation. After the debut the company’s market value exceeded 300 billion yuan, roughly 50 times its 2025 net profit. At the offer price the same profit implied a multiple nearer 21 (our arithmetic). The same shares were a very different proposition at the first close.
- Capital. The 24.5 billion yuan raised makes this the largest IPO in Shenzhen’s history. The use of proceeds is the first thing to check in the listing documents.
- Financial quality. Net profit fell to 6.1 billion yuan in 2025 from 8.28 billion in 2023, as solar tariffs fell 24% and wind tariffs 22%.
- Governance. The company is a spin-off controlled by Hong Kong-listed China Resources Power, part of the state-owned China Resources Group. As STAR showed on the JSE in 2017, a minority shareholder’s risk includes the controlling shareholder’s.
- Growth burden. At 50 times earnings, the price assumes a recovery in profits that falling tariffs have so far worked against.
Dangote Petroleum Refinery: Africa’s biggest IPO, still open
Tier: Watch. The offer of 4.1 billion new shares at ₦525 opened on 14 September and closes on 13 October, with up to 30% more shares if it is oversubscribed. Listing on the Nigerian Exchange is targeted for November.
- Valuation. Institutions bought shares in a July private placement at $0.35, a price that implied about $40 billion. The public offer at ₦525 is about $0.40 a share, roughly 14% more than the placement investors paid, and implies nearer $46 billion (our arithmetic). Ask why the public pays more than the institutions did two months earlier.
- Capital. The shares are new. According to InsidEntity’s review of the offer, net proceeds of about ₦2.11 trillion go to growth capital expenditure on the expansion. Also check how a $3.99 billion loan from Dangote Industries is to be settled.
- Financial quality. The refinery reported first-half 2026 revenue of ₦19.13 trillion ($13.91 billion), up 121%, and net profit of $1.82 billion. These are half-year figures. InsidEntity’s FSR uses only full-year audited statements, and the refinery has been producing petrol only since September 2024, so its audited record as a full operating refinery is short.
- Governance. InsidEntity’s CRR is 3.70, Good: Director Capacity 2.78, Auditor Independence 4.00 (Deloitte, five financial years), Shareholder Influence 3.00. Director Independence shows 5.00 as a default until the directors’ signed questionnaires are in. The Dangote group keeps the vast majority of shares, so read the related-party section closely. See Dangote’s full record on InsidEntity, or our analysis of the offer.
- Growth burden. The price assumes the expansion is delivered and that margins hold as the refinery’s tax position changes from 2028.
Airtel Money: a sale by the owners, not a raise for the business
Tier: Watch. Airtel Africa’s mobile-money arm announced its London listing on 23 September, after delaying it from the first half of the year. The prospectus is expected in early October and pricing in mid-October.
- Valuation. The offer is reported to raise about $800 million at a market value of $8 billion to $9 billion, about 6 to 7 times revenue. That is the 5-10x bucket, where large US IPOs trailed the market by 8.0 points over three years: a far lighter burden than SpaceX or Cerebras carry.
- Capital. Existing shareholders are selling, and the business raises no new capital. That is not automatically a warning sign: Boxer’s 2024 JSE listing also paid a parent, and its shares did well. The question is whether the price leaves value for the new buyer.
- Financial quality. Revenue was $1.3 billion in the year to March 2026, with an EBITDA margin of about 50%, 53 million monthly active users and $213 billion of transactions processed. Ask for net profit and operating cash flow over three audited years, not EBITDA alone, and for how much of the revenue is earned in currencies that have weakened against the dollar.
- Governance. The parent group holds about 78%, with TPG, Mastercard, the Qatar Investment Authority and Chimetech Holding as minority shareholders. The parent will keep control after listing, so the independence of the board and the terms of dealings with Airtel Africa matter. See the parent, Airtel Africa, on InsidEntity.
- Growth burden. The customer base has grown about 20% a year since 2018 and processed value about 33% a year. The price needs that growth to continue.
OpenAI and Anthropic: what to read first when the prospectus arrives
Tier: Close Watch. OpenAI and Anthropic have both submitted confidential draft registrations to the SEC. Neither has published a prospectus, a price or a share count, so none of the five questions can be answered yet. When the documents arrive, these are the pages to read first.
- Valuation. Price against revenue, not against the story. The evidence is clearest at the extreme: large IPOs priced above 40 times sales lost 44.8% on average over three years.
- Capital. How many of the shares are new, and how many are sold by early investors and employees.
- Financial quality. Cash flow after the cost of computing power, and any long-term commitments to buy it. Cerebras showed how far reported profit and cash can diverge.
- Governance. Both companies have unusual structures for a listed company: OpenAI is a public benefit corporation controlled by a foundation, and Anthropic is a public benefit corporation with a Long-Term Benefit Trust that elects part of its board. Read exactly what public shareholders can and cannot decide.
- Growth burden. What the offer price assumes about revenue growth, and what happens to the valuation if that growth slows.
What we don’t know yet
- How any of these listings will perform. We do not estimate it.
- Final terms for Dangote (the offer closes on 13 October) and Airtel Money (pricing expected in mid-October).
- Anything about 2026 IPOs as a cohort. The academic dataset adds them with its 2027 edition; three-year outcomes can only be measured after December 2029.
This page will be updated when Dangote and Airtel Money price, and when OpenAI or Anthropic publish a prospectus. Each update will carry its date.
The bottom line
The five questions give different answers for each of this year’s biggest names, and that is the point. Airtel Money carries a modest valuation burden, but its owners take all the money. SpaceX and Cerebras put the money into the company, but at multiples the historical record treats as extreme, with control concentrated in a few hands. China Resources New Energy is a reminder that the same shares can be a reasonable buy at the offer price and an expensive one at the first close. Dangote is still open, and its answers depend on documents investors can read before 13 October.
None of this says which of these shares will do well. It says where each one’s risk sits, and those answers are in the prospectus before the first trade.
Read the prospectus. Then check the board’s record on InsidEntity.
Independent research. Not financial or investment advice. No allegation of wrongdoing is made against any individual or entity named.
Know more. Risk less. Decide better.
Sources: Valuation multiples and returns from a given entry price are InsidEntity arithmetic on the cited figures.
- Renaissance Capital, 2026 IPO market stats, accessed 26 September 2026
- StockAnalysis, all 2026 IPOs, as of 22 September 2026
- Ritter, J. R., Initial Public Offerings: Updated Statistics, University of Florida, 14 September 2026 (Tables 1, 1a, 5, 13a, 16b, 18d, 19, 21)
- InsidEntity, SpaceX: first filing turns estimate into a price, 6 August 2026
- InsidEntity, Africa’s biggest IPO opens Monday. Its governance rating is 3.70, Good, and the pillar still open is the board, 11 September 2026
- InsidEntity, The IPO Test: Why a Great First Day Can Be a Bad Investment, 26 September 2026
- Cerebras, S-1/A registration statement, May 2026; and Mostly Metrics, Cerebras S-1 breakdown, 12 May 2026
- SpaceX, free writing prospectus, SEC, June 2026
- Dangote Petroleum Refinery, offer documents, September 2026; H1 2026 results as reported by Legit.ng, 8 September 2026
- Anthropic, announcement of confidential draft S-1 submission, 1 June 2026; OpenAI confidential draft registration, June 2026
One Response