OpenAI's $5.7B Quarter: Record Revenue, Record Losses, and the Codex Effect

OpenAI generated approximately $5.7 billion in revenue during the first quarter of 2026, according to a report from The Information, cementing its position as the highest-grossing AI company in the world—if not yet the most profitable one. The figure puts Sam Altman’s company roughly $1 billion ahead of Anthropic, which generated an estimated $4.8 billion over the same three-month period, and represents a company now running at close to $2 billion in monthly revenue.

The numbers are staggering by any startup benchmark. But they arrive alongside a more sobering counterpoint: OpenAI’s own internal projections show it burning through roughly $14 billion in net losses in 2026, against total annual spending approaching $22 billion. Record-setting revenue and record-setting losses are, for now, the same story.

The Codex Effect

If one product line deserves credit for the quarter’s momentum, it’s Codex. OpenAI’s flagship coding agent, which left limited access earlier this year to reach a broader enterprise audience, has become one of the company’s fastest-growing revenue contributors. Enterprise software teams at major banks, healthcare systems, and technology companies have embedded Codex into core engineering workflows, and the resulting contract values are translating into meaningful ARR lift.

OpenAI has not disclosed Codex-specific revenue, but analysts tracking enterprise deployment patterns estimate the product family—including the Codex CLI, Codex in ChatGPT, and the API-accessible o-series reasoning models—now accounts for a meaningful share of the enterprise bookings surge that pushed commercial revenue above $2.3 billion in the quarter.

Enterprise now represents more than 40% of OpenAI’s total revenue mix, up from roughly a quarter of the business a year ago, and the company is guiding internally toward commercial and consumer segments reaching parity before the end of 2026.

Consumer Base Holds, Quietly Expands

On the consumer side, ChatGPT reached 55 million paying subscribers in Q1, up from approximately 47 million at the close of 2025. Weekly active user figures hovered around 905 million globally—a number that reinforces ChatGPT’s position as one of the most-used software products on earth, even as the app faces saturation pressure in core markets like the United States and Western Europe.

The ChatGPT advertising pilot, quietly rolled out in early 2026 with a self-serve Ads Manager for select partners, has not yet contributed materially to revenue. OpenAI’s advertising ambitions are a longer-term bet that the company is reluctant to foreground, given the reputational sensitivities around monetizing a product many users treat as a trusted information companion. But the infrastructure is being built, and the Q2 numbers will offer a cleaner read on whether ad-supported tiers can meaningfully change the unit economics.

The Paradox of Profitable Growth Without Profit

The financial picture OpenAI presents to investors is one of the more unusual in Silicon Valley history: a company with a nine-figure monthly revenue run rate that is simultaneously burning billions every quarter. The gap between revenue and total cost reflects the extraordinary expense of training frontier models, operating the inference infrastructure that powers 900 million weekly users, and staffing the research and safety teams that underpin OpenAI’s credibility as a responsible AI developer.

Compute costs alone run into the billions annually, with OpenAI relying on a combination of Microsoft Azure capacity, the recently announced Colossus supercluster operated through a deal with SpaceX, and a growing owned-infrastructure footprint funded in part by the $122 billion fundraise completed in March. That round, which valued OpenAI at $852 billion and brought in Amazon, SoftBank, and Nvidia as anchor investors, was designed explicitly to fund the capital expenditure needed to sustain frontier model development through a period when revenue—though growing fast—still trails costs.

OpenAI’s own multi-year projections, which have circulated in investor materials, show cumulative losses in the range of $44 billion between 2023 and 2028, with the company turning cash-flow positive around 2029–2030 as revenue climbs toward $100 billion annually. The 2026 figure of $14 billion in expected losses implies roughly $1.60 spent for every dollar earned—a ratio that has improved from prior years but remains a significant structural challenge on the road to a planned IPO.

Anthropic Closes Fast, Aims Higher

The most notable context for OpenAI’s Q1 numbers is what they reveal about the competitive dynamic with Anthropic. The $4.8 billion Anthropic generated in Q1 was itself a historic figure—but the more significant data point is what Anthropic is projecting for Q2: $10.9 billion in revenue, roughly double its Q1 run rate. If that guidance holds, Anthropic would not only narrow the gap with OpenAI but potentially surpass it on quarterly revenue within the year.

Anthropic crossed into operating profitability for the first time in Q2 2026, a milestone that distinguishes it structurally from OpenAI at this stage of the AI arms race. The contrast is deliberate: Anthropic has consistently positioned itself as the company that can scale responsibly, and turning a profit—even a narrow one—is a credibility signal the company is not shy about amplifying.

For OpenAI, the competitive pressure is welcome in one sense: it validates the market they’ve collectively created. But it also forces a harder look at whether the company’s aggressive spending on frontier model development, hardware acquisition, and new product lines like an AI-native device (slated for 2027) can be justified before the IPO that Altman has telegraphed for late 2026 or early 2027.

What the Numbers Mean for the IPO

OpenAI filed a confidential S-1 with the SEC in late April, with a public offering expected no earlier than September 2026. The $5.7 billion Q1 revenue figure is the single most important data point the company can offer prospective public-market investors, and it lands well. Annualized, it implies a run rate above $22 billion—roughly in line with internal targets—and the trajectory toward $25 billion in full-year revenue is plausible if enterprise momentum holds.

The harder sell is the loss figure. Public markets have historically been forgiving of growth-stage losses in hypergrowth sectors, but the combination of OpenAI’s size (no longer a startup by any reasonable definition), its burn rate, and the uncertainty around when the AI infrastructure investment cycle peaks will require careful positioning.

What Altman and CFO Sarah Friar will need to argue convincingly is that the path from $14 billion in losses to profitability is well-understood, time-bounded, and not dependent on hypothetical future breakthroughs. The Q1 revenue report is a good first page of that argument. The rest of the story—Codex at scale, enterprise penetration, AI device upside—is still being written.