OpenAI's third-quarter numbers show a company growing fast and spending faster. According to internal figures the company presented to its investors, revenue climbed nearly 50% from the second quarter to reach $10 billion. Over the same period, the operating loss grew from $10.1 billion to $10.9 billion. That combination, strong top-line acceleration alongside a loss that outpaces revenue, is the central tension in OpenAI's financial picture as the company works toward a public listing planned for the first half of next year. The contrast with Anthropic, which reported $17 billion in quarterly revenue and a loss of just over $500 million, sharpens the question of what each company's cost structure signals to investors and enterprise buyers.

The headline figures

The quarter marks a sharp turnaround in growth rate. In the second quarter, OpenAI's sequential revenue growth had slowed to 17%, a figure that raised concern internally and among investors about whether adoption was flattening. The jump back to roughly 50% growth in the third quarter eased those worries.

The loss side moved in the opposite direction in absolute terms, though not in relative terms. Adjusted operating margin improved from negative 150% to negative 109%. In plain language, OpenAI still lost more than it earned, but the loss shrank as a share of revenue. That pattern suggests some operating leverage: revenue is growing faster than costs, even though costs are still enormous.

Metric Q2 Q3
Revenue growth (quarter over quarter) 17% nearly 50%
Revenue — $10 billion
Operating loss $10.1 billion $10.9 billion
Adjusted operating margin -150% -109%

One accounting detail matters for anyone reading these numbers. The $10.9 billion operating loss excludes stock-based compensation, which is estimated at roughly $2.6 billion for the quarter. Stock-based compensation is a non-cash expense, meaning the company pays employees in equity rather than cash, so analysts often strip it out to see underlying cash burn. Even on that basis, the loss remains close to $11 billion for a single quarter.

Finance office at night with a steep green column beside a taller red column and a whiteboard chart

▲ Revenue growth versus widening operating loss

What drove the revenue jump

Two forces appear to explain the acceleration: new model releases that gained strong user traction, and enterprise sales.

Enterprise overtakes consumer

OpenAI built its early reputation on consumer products, but the majority of consumer users remain on free tiers that generate no subscription revenue. Enterprise customers, by contrast, pay for software, API access and tools. Greg Brockman, an OpenAI executive, has said that enterprise sales now account for the majority of the company's total revenue, and that enterprise revenue grew 50% in the third quarter alone.

Codex, OpenAI's code generation product, stands out as an example of this shift. The company devoted heavy development resources to it, and that investment translated into revenue from enterprise software customers. The lesson appears to be that products aimed at paying professional users convert to revenue more reliably than broad consumer reach.

Advertising reaches $1 billion run rate

OpenAI's advertising business has reached an annualized run rate of $1 billion. A run rate takes the most recent period's revenue and projects it over a full year, so it describes current pace rather than money already collected. The advertising effort is growing quickly, but whether it will hit the company's original year-end targets remains uncertain.

Beyond ads, OpenAI continues to test several revenue channels at once: subscriptions, enterprise software, APIs and advertising. The unresolved question is how the company will extract value from the hundreds of millions of users who use its products for free. That question sits at the heart of the profitability debate.

Why losses keep growing

If revenue is rising nearly 50% a quarter, why did the loss grow at all? The answer lies in OpenAI's business model, which is broader and more sprawling than Anthropic's. OpenAI runs consumer apps, enterprise products, developer APIs, an advertising business and a research organization, and it is committing massive capital to infrastructure. Leadership frames these outlays as long-term strategic investments toward artificial general intelligence, the goal of building AI systems that match or exceed human capability across most tasks.

Those commitments show up as upfront spending on computing infrastructure and research well before the revenue they are meant to generate arrives. That timing mismatch is why the loss can widen in dollar terms while the margin improves in percentage terms.

Vast data center hall with rows of server racks, overhead cooling pipes and technicians unpacking new racks from crates

▲ Upfront infrastructure spending behind the loss

The Anthropic comparison

Anthropic's third-quarter results present a very different profile. The company brought in $17 billion in revenue, well ahead of OpenAI's $10 billion, and reported an operating loss of just over $500 million.

Company Q3 revenue Q3 operating loss
OpenAI $10 billion $10.9 billion (excluding about $2.6 billion in stock-based compensation)
Anthropic $17 billion just over $500 million

A caveat applies before drawing conclusions. Anthropic includes pass-through revenue from cloud partners in its reported totals. Pass-through revenue is money that flows through a company's books on its way to another party, in this case cloud providers that host and sell Anthropic's models. Booking that revenue inflates the top line relative to a company that reports only what it keeps. So the raw revenue gap likely overstates the difference in what each company actually earns.

Even after allowing for that difference, Anthropic appears to have opened a noticeable gap on both top-line figures and cost containment. A loss of roughly half a billion dollars against a loss near $11 billion is a difference of scale, not of accounting treatment alone. The most plausible explanation is the one already noted: Anthropic runs a narrower business, while OpenAI is funding a much wider set of products and infrastructure bets at the same time.

Funding the gap

A loss of this size requires constant fresh capital, and OpenAI has structured its latest financing accordingly. The company originally sought $30 billion, then reduced the target to $20 billion and split it into two parts.

  • First tranche: $11 billion closed at an implied valuation of $1.25 trillion
  • Second tranche: $9 billion expected to close within the coming weeks
  • Valuation: $1.25 trillion appears to be holding across both tranches

The valuation holding steady between the tranches suggests investor appetite has not weakened despite the loss figures. The company is also tracking toward an initial public offering in the first half of next year. That timeline is not guaranteed. If public equity markets soften, or if institutional investors intensify their scrutiny of the path to profitability, the schedule could shift.

What the numbers signal

For investors, the quarter offers two readings that are both true. Growth re-accelerated sharply, and enterprise adoption appears to be the engine that will carry revenue forward. At the same time, a loss of nearly $11 billion in a single quarter is an extraordinary burn rate, and it raises real questions about how long the model can run before margins turn. The improving adjusted margin is the strongest argument that the trajectory bends toward profitability; the growing absolute loss is the strongest argument that it has not bent yet.

For enterprise customers, the signal is more practical. Both companies are investing heavily to serve business buyers, and the fact that enterprise now drives the majority of OpenAI's revenue suggests that segment will shape product priorities. Anthropic's leaner loss profile may read as a sign of discipline, though its reported revenue includes cloud pass-through that complicates a direct comparison.

How to read AI company financials

The third-quarter figures offer a useful checklist for anyone evaluating generative AI businesses:

  1. Track enterprise adoption and API usage rather than free user counts. Free users cost money to serve and have not yet converted into revenue at scale.
  2. Check how revenue is recognized before comparing competitors, especially whether cloud partner pass-through sales are booked as company revenue.
  3. Separate core operating burn from non-cash expenses such as stock-based compensation to understand actual cash needs and runway.
  4. Watch the trend in margin, not just the dollar loss. A loss that grows in dollars but shrinks as a share of revenue tells a different story than one that grows on both measures.

OpenAI's third quarter shows a company that has found its growth engine in enterprise sales while still paying the full upfront cost of its infrastructure ambitions. With $11 billion of a $20 billion round already closed at a $1.25 trillion valuation and an IPO on the horizon, the next few quarters will reveal whether revenue growth can finally outrun the spending that supports it. Readers following the sector should watch the second funding tranche, the enterprise growth rate and the advertising run rate against year-end targets as the clearest early indicators.