The number and where it came from
Bloomberg reported that OpenAI's annualized revenue run rate surpassed $40 billion as of August 2026, citing people familiar with the company's internal figures. The rate climbed more than 20 percent in July alone, month over month, which on an annualized basis works out to roughly double where OpenAI stood at the end of 2025. Cofounder Greg Brockman confirmed the trajectory internally, noting the acceleration explicitly rather than leaving it to outside reporting.
The growth is broad-based rather than concentrated in one product. Bloomberg's sourcing points to AI coding software, specifically Codex, alongside subscription sales, a newly launched advertising business, and continued growth in the core consumer ChatGPT product. OpenAI also appointed a second chief revenue officer in under a year, a staffing signal that the company is building out sales infrastructure to match the growth rather than treating it as a temporary spike.
Enterprise is now the tell, not the consumer number
The detail that matters most to a CTO reading this sits below the $40 billion headline: enterprise sales now represent over 40 percent of OpenAI's total revenue. A year ago the OpenAI story was almost entirely a ChatGPT consumer subscription story with enterprise as a smaller, slower-growing add-on. That mix has shifted meaningfully, and it lines up with what buyers have been seeing directly: more account management attention, more willingness to negotiate custom terms, and a faster cadence of enterprise-specific product launches like Ultrafast inference and dedicated agent tooling. It also means OpenAI has real financial incentive to keep investing in the parts of the product that enterprise buyers actually use, rather than treating enterprise as a side channel to its consumer app.
OpenAI also cut prices on certain models to compete against rivals, which sits awkwardly next to a revenue run rate that just doubled. The combination suggests OpenAI is trading margin on commodity workloads to win volume, while relying on premium tiers, enterprise contracts, and the new advertising line to carry profitability. That is a familiar playbook from cloud infrastructure vendors, and it means pricing pressure on OpenAI's lower tiers is likely to continue even as the company's overall revenue keeps climbing.
The IPO race nobody has officially started
OpenAI has filed confidential paperwork to go public, according to Bloomberg, though no exact timeline has surfaced. What makes this notable is the competitive context: Anthropic is reportedly further along in its own IPO process and could reach public markets first, potentially as early as fall 2026. For two labs that have spent three years positioning themselves as fundamentally different, one safety-focused and cautious, the other consumer-scale and fast-moving, a near-simultaneous race to the public markets says the financial logic driving both companies has converged even where their branding has not.
A public listing changes the information environment for every enterprise buyer. Once either company trades publicly, procurement and finance teams get access to audited financials, disclosed unit economics, and quarterly guidance that today they have to infer from leaked figures and anonymous sourcing. That is a meaningful upgrade in vendor risk assessment for a category where buyers currently sign multi-year commitments on companies that disclose almost nothing about their actual cost structure.
What the report does not say is the real risk
Bloomberg's reporting is explicit about revenue and silent on profitability. Compute costs, infrastructure lease obligations, and model training spend at OpenAI's scale are enormous and have not shrunk as fast as revenue has grown industry-wide. A revenue run rate doubling is genuinely impressive as a demand signal, and it is also a different metric from cash flow, and enterprise buyers should not conflate the two when assessing vendor durability for a five-year contract. The gap between those two metrics is exactly where every prior tech bubble has hidden its worst surprises, from telecom in 2001 to cloud infrastructure overbuilds a decade later, and it is worth naming plainly rather than assuming growth alone answers the durability question.
OpenAI's growth appears real and diversified across product lines, and the caution here is about incomplete data rather than distrust. A headline revenue figure without a matching profitability figure tells you about demand, and demand alone does not establish a vendor's staying power at current pricing over a multi-year contract. Any procurement conversation this quarter should ask directly what OpenAI, and any AI vendor citing similar growth numbers, is willing to disclose about unit economics rather than accepting the top-line figure as the whole picture.
What this means for your vendor strategy
A $40 billion run rate and over 40 percent enterprise mix are strong evidence that OpenAI intends to stay the dominant enterprise AI vendor rather than cede that ground to Anthropic or Google, and buyers should expect continued investment in enterprise features, support, and account coverage as a result. That reduces one category of vendor risk, the risk that OpenAI deprioritizes enterprise to chase consumer growth, which was a legitimate worry as recently as a year ago.
It does not reduce pricing risk or lock-in risk. As OpenAI approaches a public listing, expect more pressure to show margin expansion, which historically means price increases on premium tiers even as commodity tiers get cheaper to defend market share. Build contract terms now that protect you from both directions: price caps on the models you depend on most, and the technical flexibility to route workloads elsewhere if OpenAI's post-IPO pricing moves against you.
