The news
OpenAI is on track to generate more than $40 billion in annualized revenue. That figure is roughly double the company’s run rate at the end of 2025. Bloomberg Technology reported the projection on August 14, 2026, citing sources familiar with the company’s performance.
The growth comes from three main areas: paying ChatGPT users, broader enterprise adoption, and usage of the Codex coding assistant. Bloomberg’s Rachel Metz highlighted these drivers during an appearance on “Bloomberg Tech” with Tim Stenovec.
Context
At the close of 2025, OpenAI operated at a lower revenue baseline. The current trajectory represents a sharp acceleration within roughly eight months. Enterprise customers and individual subscribers are both contributing to the increase.
Metz noted that the company continues to face heavy infrastructure spending. The cost of computing power required to train and serve models remains the dominant expense and a central operational challenge.
Details
Paying ChatGPT users form one clear growth vector. The number of subscribers and the average revenue per user have both risen, according to the Bloomberg reporting. Enterprise deals add a second layer, with organizations integrating OpenAI models into internal workflows at scale.
Codex, the coding assistant, supplies a third source of revenue. Developers and companies are adopting it for code generation and completion tasks, increasing overall usage and billed compute. Metz described these three elements together as the primary reasons the annualized run rate is climbing so quickly.
The same report flags the persistent cost pressure. Training and inference workloads demand large clusters of specialized hardware, and the associated electricity and hardware expenses continue to rise with usage. No offsetting figures on margins or profitability were provided in the segment.
Why it matters
For software engineers and technical founders evaluating platform commitments, the revenue trajectory signals that OpenAI is scaling its commercial footprint faster than most competitors. The doubling of annualized revenue in under a year shows sustained willingness to pay for frontier models in both consumer and enterprise settings.
At the same time, the explicit call-out of compute costs underscores a structural constraint. Any organization building on OpenAI APIs will ultimately bear some portion of those infrastructure expenses through pricing. The combination of rapid top-line growth and heavy capital requirements suggests the company’s path to an IPO will be watched closely for evidence that revenue can outpace spending over multiple quarters.
The reported numbers also set a benchmark for other AI providers. If OpenAI sustains this pace, competitors will need to demonstrate comparable adoption curves or risk losing ground in both developer mindshare and enterprise budgets. The next public updates on user counts, enterprise wins, and infrastructure outlays will determine whether the current run rate holds or flattens.
Engineers deciding whether to standardize on OpenAI tooling now face a clearer picture of the vendor’s momentum. A $40 billion run rate implies the company has locked in recurring usage across thousands of teams, which often translates into more stable API pricing and feature roadmaps. Yet the same scale that drives revenue also locks in massive fixed costs for GPUs and power, leaving less room for price cuts even if rivals emerge with cheaper alternatives.
Founders planning multi-year product bets must weigh this against the possibility that OpenAI’s margins stay thin for longer than expected. Heavy infrastructure outlays mean continued reliance on large funding rounds or debt before an IPO, which can shift priorities toward high-margin enterprise features over the experimental tools many developers prefer. The Bloomberg segment does not address whether OpenAI has narrowed its losses, so the durability of the growth story remains an open question for anyone whose roadmap depends on the company’s continued investment capacity.
---
Sources:
{"word_count": 612, "sources_used": 1}
No comments yet