Difference in OpenAI and Anthropic disclosures caused tech stocks to plunge

The difference in approaches to revenue recognition between OpenAI and Anthropic has caused confusion among investors and sparked a sell-off in U.S. technology stocks. It turned out that the annual revenue run rate of the ChatGPT developer as of September was about $50 billion, while the market had been using a figure of almost $70 billion.

This was reported on October 9 by Bloomberg, Financial Times, and Axios.

According to sources familiar with OpenAI's financials, the company reached an annual revenue run rate of approximately $50 billion by the end of September. That's $20 billion less than the figure widely circulated in late September.

However, the discrepancy does not mean that OpenAI lost $20 billion in revenue. The main reason lies in the different methods of accounting for sales through cloud partners used by OpenAI and its competitor Anthropic.

Where the $20 billion difference came from

As Axios discovered, the earlier estimate of $70 billion was calculated to make a more direct comparison of OpenAI's financial results with Anthropic's.

Companies use different approaches to recognizing revenue from products sold through third-party cloud platforms.

Anthropic includes the full price of such sales in revenue and records the cloud provider's commission as an expense. OpenAI, under certain partnership agreements, recognizes only its own share of the payment received.

For example, if a customer buys an AI service through a cloud platform for $100, Anthropic might record the full $100 as revenue, and then separately account for the partner commission as an expense. OpenAI, under a similar deal structure, might recognize only the amount that directly accrues to the company.

According to accounting expert Francine McKenna, as she explained to Axios, both approaches comply with U.S. GAAP standards. The choice of method depends, among other things, on who controls the customer relationship and is responsible for delivering the product.

Investors attempted to translate OpenAI's figures to Anthropic's methodology in order to compare competitors on a like-for-like basis. The resulting estimate of around $70 billion subsequently became widely used as OpenAI's annual revenue run rate.

After new data emerged, it became clear that comparing financial results of the two AI developers without adjusting for methodology could present a distorted picture of their financial performance.

Financial figures triggered a tech stock sell-off

News of a lower revenue estimate for OpenAI negatively affected investor sentiment, as the company is considered one of the largest buyers of AI computing capacity.

In trading on October 8, Nvidia fell 2.9%, AMD by 3.9%, Micron by 4.8%, Broadcom by 4.3%. Oracle's stock also came under pressure.

Investors worried that the difference in financial estimates could indicate a slower pace of AI commercialization than previously thought.

Later, analysts pointed out that the discrepancy is primarily about methodology, not a confirmed decline in demand for OpenAI's products.

The following day, some tech stocks began to recover from the initial negative market reaction.

OpenAI expects to increase revenue by year-end

Despite the confusion, the company continues to grow sales rapidly. According to Bloomberg, OpenAI expects to reach or exceed an annual revenue run rate of $70 billion by the end of 2026. The main driver of further growth is expected to be the enterprise segment.

Thus, one must distinguish between two figures: around $50 billion as of the end of September, and the expected $70 billion by the end of December. The latter is a forecast of future growth, not a confirmed result for September.

It is important to note that these figures refer to the annualized revenue run rate, a metric calculated by extrapolating sales from a shorter period to a full year. It is not the actual revenue over the trailing twelve months.

Amid the business growth, OpenAI is also in talks to raise at least $30 billion in new capital. As Bloomberg reported, the preliminary valuation being discussed is about $1.4 trillion pre-money.

The situation with OpenAI's and Anthropic's financials demonstrates how important a unified approach to measuring AI developers' revenue is becoming for investors, especially against the backdrop of massive investments in data centers and computing infrastructure.

Sources: Bloomberg, Financial Times, Axios, MarketWatch.