Why a Frontier AI IPO Valuation Cut Is Not the Same as Lower Proceeds

A frontier artificial intelligence company recently updated its listing documents and plans to list on the U.S. capital market in November 2026. Under the disclosed terms, it intends to issue about 825 million new shares at $70 to $78 per share, implying maximum proceeds of about $64.35 billion. Some existing shareholders also plan to sell shares, but the proceeds from those sales belong to the selling shareholders, not the company.

Investor discussions suggest that valuation expectations have fallen from earlier levels. The market previously discussed a target valuation of about $2 trillion. More recent expectations are closer to $1.3 trillion to $1.45 trillion, with some descriptions placing the central valuation at about $1.4 trillion. The company has also secured about $26 billion in private subscription commitments, mainly from participants across chips, manufacturing, software, and infrastructure.

“Offer price,” “proceeds,” “secondary-sale proceeds,” and “company valuation” are not the same concept. Distinguishing them is the first step in judging the listing value of an AI company.

1. Issue size is not company valuation

IPO proceeds from newly issued shares are usually determined by two factors:

Proceeds from new shares = number of new shares × offer price

Using the upper end of the disclosed range:

825 million shares × $78 per share ≈ $64.35 billion

That money goes into the company’s account and is typically used for research and development, infrastructure, debt repayment, working capital, and general corporate purposes.

Company valuation is usually calculated on a fully diluted share count:

Fully diluted valuation = offer price × total fully diluted shares

Fully diluted shares include outstanding common stock and may also include employee options, restricted stock, convertible securities, warrants, and other instruments that could dilute shareholders. An IPO can therefore raise only tens of billions of dollars while the company is valued in the trillions. The two figures are not contradictory.

When reading a prospectus, do not infer the listing percentage by simply dividing proceeds by valuation, and do not look only at the number of new shares. A more reliable approach is to check basic share capital, preferred-share conversion arrangements, employee incentive plans, and securities that could dilute ownership.

2. Secondary sales do not directly improve company cash flow

In this transaction, some existing shareholders plan to sell Class A common stock. A secondary sale means early investors, employees, or other shareholders transfer shares they already hold to public-market investors.

This can have three effects.

First, existing shareholders can obtain partial liquidity. At a young AI company whose valuation has risen quickly, early holders may have held unlisted shares for a long time. An IPO gives them a relatively clear exit route.

Second, the company does not directly receive the proceeds. Unless the offering documents provide a special arrangement, that money belongs to the selling shareholders, not to the company’s operating funds. Investors should not count secondary-sale proceeds as cash available to the business.

Third, a high proportion of secondary shares can affect market perception. If a large share of the initial offering comes from existing holders, the market may ask whether early investors are cashing out in concentration. If new shares make up a larger share, the company raises more capital, but existing shareholders face more dilution.

When reading a listing plan, separate “shares newly issued by the company” from “shares sold by existing shareholders.” Do not treat the two transactions as the same.

3. What a $26 billion cornerstone commitment can and cannot do

The company has signed private subscription agreements with multiple industry-chain participants. Those investors will subscribe, in aggregate, for about $26 billion of new shares at a price no higher than, or equal to, the IPO offer price. Markets often call this type of arrangement a cornerstone investment or strategic placement.

From a financing perspective, cornerstone capital can provide some stability:

• It locks in part of the funding before the public offering and reduces fundraising uncertainty.
• It signals that industry capital recognizes the company’s long-term value.
• It strengthens capital ties with chips, manufacturing, data centers, and the software ecosystem.
• When markets are volatile, it can provide a reference range for pricing.

A cornerstone investment is not unconditional long-term support. Investors still need to examine lockups, subscription price, exit arrangements, related-party transactions, voting design, and whether the subscription carries special rights.

Industry investors taking part may also indicate that the company will continue to need computing power, storage, networking equipment, and data-center resources. For an AI company, capital is not used only for algorithm research. Training and inference infrastructure also require substantial funding. Cornerstone capital can ease near-term funding pressure, but it does not by itself prove that the company has a stable profit model.

4. What a cut from $2 trillion means

A move in expected valuation from about $2 trillion to $1.3 trillion–$1.45 trillion does not necessarily mean the business has deteriorated. It can also mean that pricing is moving back toward a more rational level.

Valuations of frontier AI companies are usually affected by several factors at once.

Sustainability of revenue growth

The market looks not only at current revenue, but also at whether it comes from long-term contracts, ongoing subscriptions, API usage, enterprise solutions, or other repeatable business. If revenue depends mainly on a few large customers or short trials, the valuation is likely to be less stable.

Compute cost and gross margin

Training and inference require large amounts of computing resources. As the user base grows, higher revenue does not necessarily mean profits grow at the same pace. If inference cost stays above what customers pay, the company can face a pattern in which revenue rises while costs rise even faster.

Investors therefore usually focus on:

Cost per unit of revenue
Inference cost per request
Utilization of computing resources
Model-training investment
Infrastructure depreciation

Intensity of R&D spending

Frontier-model companies must keep investing in data, algorithms, engineering talent, and computing resources. Heavy R&D spending can help maintain a technical lead, but it also suppresses near-term profit. Whether a high valuation can hold depends on whether technical advantage can be turned into stable commercial revenue.

Competition and product substitution

AI models iterate quickly, so a technical lead may last only for a period. A company needs to show strong model capability, a developer ecosystem, customer stickiness, and a closed loop of data feedback. Otherwise, a high valuation is exposed to new technical paths and competing products.

5. Why fully diluted valuation matters

During an IPO, media or market participants sometimes value a company by multiplying the offer price by the current number of shares in circulation. That can understate future dilution.

A fully diluted count needs to include:

1. Common shares already issued.
2. Employee stock options and restricted stock.
3. Shares issued on conversion of convertible preferred stock.
4. Securities that have not been exercised but could dilute shareholders.
5. Shares that future incentive plans may add.

If employee equity incentives are large, the post-listing share count can be significantly higher than the headline figure. Even if the price per share stays the same, the company’s overall valuation can rise because there are more shares. For common shareholders, ownership can be diluted.

That is why the same company can be described with three different numbers: an offer-price valuation, post-listing market value, and fully diluted valuation. Investors should confirm the share-count basis and the point in time behind each figure.

6. Metrics that matter in an AI listing

For a company valued this highly, revenue or the amount raised is not enough. Judgment should cover several layers.

Operations

Focus on customer count, customer concentration, contract length, renewal rates, revenue-recognition methods, and the revenue mix across business lines. In particular, separate one-time project revenue from repeatable revenue.

Technology

Look at training and inference efficiency, service stability, data-governance capability, the R&D iteration cycle, and dependence on a single infrastructure supplier. Technical metrics do not need to chase a single peak result. What matters more is whether they can support delivery at scale.

Finance

Focus on gross margin, selling expenses, R&D expenses, cash-burn rate, and future capital expenditure. For a company that is not yet consistently profitable, how long cash reserves can last is an important basis for judging whether more financing is necessary.

Governance

Review voting arrangements across share classes, founder-team control, related-party transactions, lockups, and potential selling pressure after listing. Special voting structures can help preserve a long-term strategy, but they also reduce ordinary shareholders’ influence over major decisions.

7. How to view a high-valuation IPO rationally

A high valuation is the market’s combined price for future cash flow, technical position, and industry opportunity. It is not a simple reward for current profit. A strong technical base and fast growth potential can justify a high valuation for an AI company. If the valuation rests mainly on distant expectations, market swings are also likely to be larger.

For investors, the most reliable order of analysis is:

Confirm the share-count basis
Separate new shares from secondary shares
Check the use of proceeds
Analyze revenue quality and cost structure
Then judge whether valuation matches growth expectations

For company managers, listing is not the end of financing. After the public offering, the company still has to keep showing three things: that technology spending can become commercial value, that infrastructure costs can be controlled, and that new shareholders can receive long-term returns that match the risk.

Overall, the valuation cut reflects capital markets reassessing a model of high growth and high consumption. About $26 billion of industry-capital subscriptions improves financing certainty, but it cannot replace the company’s own ability to earn profits. What will still decide post-listing performance is the quality of revenue growth, control of compute costs, the efficiency of technical iteration, the degree of equity dilution, and the long-term stability of the governance structure.