[Sokbohyo [unclear caption]] OpenAI revenue shock: signal or noise?
- What to look for
- The reason for the discrepancy between 700 hundred million and 500 hundred million dollars can be examined through the difference between gross and net revenue calculations and an analogy involving finished products and raw materials.
- Why it matters
- When stock prices plunge on revenue reports, this helps distinguish noise from signal by examining the calculation, timing, and whether the news was already priced in, rather than just the number itself.
AI summary
The speaker reports that semiconductor stocks plunged after a Financial Times report indicated that OpenAI’s annualized revenue was around 500 hundred million dollars rather than the previously reported 700 hundred million dollars. They explain that the market had expected a revenue trajectory toward 700 hundred million dollars and was disappointed when that was not the case, which was reflected in declines of 3% for Nvidia and 5% for Micron. They say the key issue is the difference in revenue recognition: Anthropic records the total amount paid by customers as revenue and accounts for cloud costs separately, whereas OpenAI records net revenue after subtracting amounts payable to Microsoft and others. They interpret this to mean that, if converted to a gross amount on the same basis, the figure of 700 hundred million dollars is not necessarily wrong either, and the direction of the fundamentals has not changed. An actual revenue decline would indeed be a signal, but they judge it slightly more likely to be noise, citing three factors: differences in the calculation, developments after the launch of Dachi [unclear caption] on 9/29 not being reflected, and the stock price already having priced this in. All directional outlooks are the speaker’s conditional judgments, not definitive predictions of short-term prices.