What will OpenAI's annualized recurring revenue (run-rate revenue) be at the end of December 2028?

rationale:

OpenAI's annualized revenue run-rate sits at roughly $50B as of late September 2026 3 sources, up remarkably from about $20B at the end of 2025 sacra.com. The company targets a run-rate of at least $70B by the end of 2026, driven heavily by enterprise adoption, which saw over 100% quarter-over-quarter growth in Q3 3 sources. Internal presentations from mid-2026 outline a steep trajectory, projecting calendar revenue rising from $36B in 2026 to $350B in 2030 2 sources. Backing out the cumulative $840B revenue target over this period implies calendar revenues of about $85B in 2027 and $140–150B in 2028, translating to a December 2028 exit run-rate in the $175–200B range if the internal plan holds perfectly.

The median forecast of roughly $156B applies a mild but necessary discount to these aggressive internal targets while aligning the long-term projection with the expected 2026 and 2027 revenue run-rates and the natural deceleration in growth. Starting from a $65–70B run-rate at the end of 2026, applying typical software deceleration—roughly 1.6–1.8x growth in 2027 and 1.35–1.45x in 2028—lands in the vicinity of the $150–165B range, supporting a $156B midpoint. The central 50% interval of $115B to $213B captures substantial uncertainty around how quickly that deceleration occurs. This assumes continued expansion across API usage, coding tools, subscriptions, and a nascent advertising business that reached a $1B run-rate by August 2026 and is projected to hit $25B in 2028 2 sources. However, maintaining hypergrowth becomes structurally harder at this scale, especially with physical compute availability operating as a binding constraint on revenue generation openai.com.

Downside risks, which anchor the 10th percentile around $86B, are dominated by capital intensity, compute bottlenecks, and competition. OpenAI projects approximately $278B in cumulative negative free cash flow through 2030, raising the possibility of funding shortfalls or forced spending reductions that cap revenue growth depthgrid.in. Furthermore, competition remains intense—Anthropic was already operating at an estimated $65B gross run-rate by mid-2026 finance.yahoo.com. We must also factor in potential safety-related setbacks, as internal fears have previously delayed model releases 2 sources, and the likelihood of a 2027 IPO 2 sources, which could force a transition to stricter public reporting metrics (e.g., Q4 revenue multiplied by four) that naturally lag a point-in-time December run-rate.

The right tail is particularly fat, with the 90th percentile extending near $285B, driven by explosive operational upside and accounting quirks. Operationally, rapid compounding of enterprise seats, widespread deployment of highly reliable AI agents, and a scaling ad network could comfortably push run-rates above $200B, with the 75th percentile at approximately $213B. Crucially, resolution risk artificially widens this upper bound: OpenAI typically reports its share of cloud-partner sales on a "net" basis, whereas competitors like Anthropic book gross channel revenue 2 sources. If credible reporting in 2028 cites a "grossed-up" figure to allow for apples-to-apples industry comparisons 2 sources, the resolved value could be inflated by roughly 1.4x over the true net run-rate, pushing the outcome into the high $200Bs.