$20 billion short: How OpenAI's revenue controversy is shaking up the AI market

Editor's note: On October 8, the Financial Times reported that OpenAI's annualized revenue was close to $50 billion as of the end of September, lower than the roughly $70 billion previously circulating in the market. After the news was published, shares of several chip companies fell. The revenue figure of a private company quickly became the focus of a repricing across the entire AI sector.

Subsequently, explanations of the statistical basis for gross versus net amounts, as well as reports on OpenAI's year-end revenue target, eased some concerns. But why the two figures differ, and whether AI commercialization revenue can support continued expansion, are two questions that need to be answered separately. Clarifying the accounting basis can correct market perceptions, but future growth still needs to be delivered through performance.

In a market commentary published on October 12, Compounding Value interpreted this selloff as an information disclosure dilemma in the AI trade: the growth expectations of chip, cloud computing, and data center companies are closely tied to a small number of model companies with limited financial disclosure; as expansion relies more on financing, revenue quality may also affect risk judgments across the entire industry chain.

This volatility is worth watching because investors are simultaneously judging two things: how fast the AI business is actually growing, and whether that growth can support the next round of computing power investment. Subsequent revenue, procurement plans, and financing conditions will determine whether the revenue controversy is merely a short-term disturbance or a more lasting valuation pressure.

The following is the original text:

A $20 billion gap has shaken AI growth expectations

On October 8, OpenAI's revenue news became the focus of pressure on the AI sector.

The Financial Times reported that information OpenAI provided to investors showed its annualized revenue was close to $50 billion as of the end of September. Previously, figures reported by Reuters and other media outlets were close to $70 billion, a difference of about $20 billion.

That day, shares of chip companies including Intel, Nvidia, Marvell, AMD, Micron, Broadcom, and Arm generally fell. Market coverage by The Wall Street Journal showed that declines in these stocks ranged from about 2% to 7%. Concerns triggered by the revenue news spread to listed companies that provide computing power and infrastructure for AI businesses.

"Annualised revenue" here refers to converting the recent revenue pace into a full-year scale, and does not equal full-year revenue already realized. Therefore, the $20 billion difference does not mean the company lost an equivalent amount of cash, nor can it be directly understood as full-year revenue being cut by $20 billion.

The market is facing a clear gap in the figures that underpinned previous growth assumptions.

From a valuation perspective, this raises a series of questions: if model companies' commercialization revenue falls short of market perceptions, can they still sustain their planned compute procurement? Does data center expansion need to be adjusted? Has future order growth for suppliers already been priced into their stocks?

These concerns can affect stock pricing even before procurement plans change. However, stock price fluctuations do not yet equate to order changes. This sell-off primarily reflects a shock to growth expectations, and whether actual demand is weakening still requires operating data to verify.

$50 billion vs. $70 billion: Definitions and timing need to be disentangled

The gap between the two figures involves two layers of easily confused issues: what basis of accounting is used for revenue, and what point in time the figures correspond to.

Compounding Value explained the earlier figure of approximately $70 billion as a gross basis that includes some partnership channel revenue, and interpreted the approximately $50 billion as a net basis. If the two figures cover different scopes, one cannot directly conclude that business revenue declined by $20 billion.

The difference between gross and net affects how partnership business revenue is presented, but net revenue is also not equivalent to profit after deducting all costs. At present, publicly available information is still insufficient to confirm that the entire discrepancy can be explained by definitional differences.

A subsequent Financial Times report continued to list OpenAI's annualized net revenue as close to $50 billion, while also disclosing that the company raised its full-year 2026 revenue forecast. This also indicates that the current annualized revenue run rate, full-year revenue forecast, and the recognition method for partnership channels need to be understood separately.

On October 9, Bloomberg reported that OpenAI expects to reach or exceed $70 billion in annualized revenue by year-end. This news was seen by the market as one of the factors behind the rebound in semiconductor and other AI-related stocks that day.

But the year-end target and the revenue level at the end of September describe two different points in time. The former depends on subsequent growth materializing and cannot be used to verify that the same scale had already been reached by the end of September.

This bout of volatility—first a decline, then a rebound—demonstrates the market's sensitivity to revenue news from model companies. Definitional explanations and growth targets can ease concerns, but they cannot replace continuous, clear, and comparable financial disclosures.

Why did OpenAI's revenue controversy transmit to chip stocks?

Chip stocks trade not only on orders already delivered, but also on expectations that customers will continue purchasing in the future. When OpenAI's revenue figures are called into question, the market may reassess: can model companies' commercialization progress support the next round of compute investment?

This explains why supplier revenue growth in the current period and pressure on stock prices can occur simultaneously. Completed procurement and deliveries can be reflected in supplier performance; customers' future spending capacity, meanwhile, affects growth expectations and valuations.

Capital expenditure (Capex) refers to the funds enterprises invest in long-term assets such as chips, servers, and data centers. These investments can translate into orders and revenue for suppliers, but for the buyer, commercial returns must be realized gradually through subsequent business operations.

Model companies need to rely on subscriptions, enterprise services, and other businesses to continuously cover computing power and operating costs. The revenue suppliers have already secured cannot alone answer how much customers can still invest in the future.

In Compounding Value's view, the reason revenue news from model companies can move the entire sector is precisely because the market treats it as a signal of end demand. When that signal is called into question, investors may re-examine the growth expectations previously assigned to suppliers.

This linkage also has its boundaries. Different suppliers have different customer structures, contract arrangements, and business portfolios. OpenAI's revenue controversy cannot directly represent demand conditions across the entire AI market. Subsequent customer procurement plans, project progress, and supplier orders are the basis for judging whether pressure has entered the operational level.

The more expansion relies on financing, the more important revenue realization becomes

Behind the revenue controversy lies another issue: the source of funding.

In Compounding Value's view, AI infrastructure buildout increasingly relying on debt financing means investors' focus will extend from revenue growth rates further to cash flow, financing costs, and debt-servicing capacity.

Revenue growth can support expansion expectations, but whether investment plans can be sustained also depends on whether funding is in place and whether operating returns can gradually materialize. If commercialization performance falls short of expectations, pressure may first show up in financing conditions, then affect capital expenditure and supplier orders.

This is a potential risk transmission path, and it does not mean that AI-related debt has already encountered repayment problems. It illustrates why the market is so sensitive to revenue figures: investors must judge not only how fast the business can still grow, but also whether the funding supporting expansion can be sustained and how much return the investment can ultimately generate.

Next, what the market needs to watch is whether revenue growth under clear standards can be realized, whether customer procurement plans continue, and whether the financing conditions required for expansion remain stable.

If revenue and orders continue to grow, the impact of this controversy on the demand outlook may weaken; only if projects are delayed, procurement is scaled back, or financing tightens will the concerns gain more direct support.

$50 billion or $70 billion affects the market's judgment of the starting point for growth; whether this revenue can support the next round of computing power investment is what determines whether this controversy will continue to depress the valuation of the AI supply chain.

Join the official Coincamps community:

X: https://x.com/coincamps

Telegram: https://t.me/coin_camps