Anthropic Passes OpenAI on Revenue by 73%, With a $2 Trillion IPO on the Horizon

The numbers that flipped the narrative

For most of the past two years, OpenAI has been treated as the assumed leader in the AI race almost by default, an assumption that took a direct hit on 18 August when the Wall Street Journal reported that OpenAI's second-quarter revenue rose 18 per cent sequentially to $6.7 billion from $5.7 billion in the first quarter. Under most circumstances, 18 per cent quarterly growth would be a strong result.

Set against Anthropic, it looked moderate. Anthropic's Q2 revenue came in at roughly $11.5 to $11.6 billion, according to the same reporting, more than double its $4.73 billion first quarter and up nearly 14-fold from $787 million a year earlier. Anthropic also posted its first quarter of positive adjusted operating income, while OpenAI's operating loss widened to $12.3 billion from $9.3 billion in the prior quarter.

Both sets of figures remain preliminary and unaudited, shared with investors ahead of anticipated public listings rather than filed as formal results. Anthropic's annualised revenue run rate has since been reported at $65 billion for July, and the company has confidentially filed for an IPO targeting a valuation of up to $2 trillion, which would rank among the largest public offerings in history.

OpenAI, valued at $852 billion after a $122 billion funding round in March, has not committed to a listing date. OpenAI chief financial officer Sarah Friar told staff at an all-hands meeting on 19 August that the company will be a public company in 2027 and possibly sooner if growth accelerates, while being direct about the comparison with her rival.

❝

"We are running our own race," Sarah told employees, according to people who attended the meeting.

What is actually driving the gap

The two companies have taken visibly different paths to revenue. Anthropic's growth has been concentrated in enterprise and developer tools, particularly Claude Code, which the company credits with pulling its run rate from roughly $9 billion to $65 billion within months. OpenAI, historically stronger with consumer ChatGPT subscriptions, has been shifting its own mix toward enterprise.

Sarah told investors on 14 August that enterprise and consumer revenue had now crossed, after starting the year at a 60-40 split favouring consumer, adding that "the majority of our revenue is now enterprise" and that the shift arrived faster than the company had forecast at the start of the year.

This repositioning matters for how AEC technology leaders should read the comparison.

A revenue lead built on coding and enterprise workflow tools, Anthropic's position, points to a company whose growth is anchored in the kind of professional software integration that construction and engineering firms are already buying into.

A revenue base still substantially weighted toward consumer subscriptions, even as OpenAI's enterprise share climbs, points to a different customer relationship and a different renewal dynamic, one built more on individual habit than on procurement contracts.

The two companies are also pricing their way through this transition differently, which is worth noting for anyone negotiating a multi-seat licence. OpenAI has leaned on aggressive price cuts to defend and rebuild its enterprise share, while Anthropic's growth so far has coincided with holding pricing largely steady on its flagship coding and enterprise products.

Neither approach has a clear advantage yet, but the divergence itself is useful information: it suggests the two companies are not simply competing on capability but on two different theories of how enterprise AI spending will settle once the current land grab slows.

The counterpoint that keeps this from being a settled story

Revenue scale is not the only signal worth tracking, and a second dataset released within days of the Q2 figures complicates any simple story of Anthropic pulling decisively ahead.

Ramp, the corporate card and expense platform, tracks spending across more than 70,000 US businesses and found that Anthropic overtook OpenAI in market share among its paying business users back in May, reaching 41 per cent against OpenAI's 39 per cent, before extending that lead to nearly 44 per cent versus almost 40 per cent by July.

Ramp economist Ara Kharazian's more recent read of the data, however, shows OpenAI growing faster than Anthropic among that same business segment so far in the third quarter, though a month remains before the quarter closes.

The reversal appears tied to pricing. OpenAI cut the price of its newest models sharply within weeks of release, slashing one model's cost by 80 per cent to 20 cents per million input tokens.

❝

TechCrunch, which first reported the Ramp data, framed the pattern bluntly: businesses are "willing to flop back and forth as each lab releases new models," a volatility that should give both companies' investors pause about how sticky enterprise AI spending really is.

For any firm treating its AI vendor choice as a long-term platform decision rather than a monthly subscription, that volatility is the more important data point than either company's headline revenue figure.

Why AEC firms should be paying attention at all

It is tempting to file quarterly AI lab earnings under general technology news and move on, but the vendor landscape underneath enterprise software is consolidating around a small number of frontier model providers faster than most procurement cycles account for.

Autodesk, Procore, Trimble and a growing list of AEC-specific platforms now embed either Anthropic's or OpenAI's models, sometimes both, inside tools firms already rely on for takeoffs, scheduling and document review.

A shift in which lab is winning enterprise share is not an abstract market story. It is a preview of which underlying model capabilities, pricing structures and data policies will show up inside next year's software renewal.

Reading this as an AEC technology buyer, not a market spectator

Neither company's quarterly figures should determine which AI tools an architecture, engineering or construction firm adopts. What the figures do reveal is a vendor landscape moving fast enough that pricing, model updates and platform positioning can shift meaningfully within a single quarter. A few implications worth carrying into procurement conversations:

Treat AI vendor selection as a portfolio decision rather than a single-platform commitment where practical, since switching costs for individual workflows are currently low and pricing is moving quickly.
Watch enterprise-specific product signals, such as coding tools, data retention policies and integration depth, over headline revenue or valuation figures when assessing which vendor is actually built for professional workflows.
Expect further price competition. OpenAI's aggressive cuts suggest neither company is treating current pricing as fixed, which should factor into any multi-year licensing conversation with a vendor or reseller.

Takeaway

Anthropic's revenue lead is real and enterprise-anchored, but Ramp's Q3 data shows OpenAI clawing back share among paying businesses within weeks, so treat any single dataset as a snapshot rather than a settled ranking.
Both companies are running unaudited, preliminary figures ahead of IPOs that could value them at a combined $2.8 trillion or more, a scale that will pull public market scrutiny onto product decisions that used to stay private.
The pricing volatility Ramp documented is the clearest warning sign for procurement teams: platform loyalty in enterprise AI currently lasts roughly as long as the next model release.
Firms with meaningful AI spend should revisit vendor agreements at each quarterly earnings cycle for the next year, since both companies are actively repricing to win share rather than settling into a stable market structure.

The vendor race behind your firm's AI stack moves week to week, not quarter to quarter, and Project Flux tracks exactly this kind of shift so your procurement conversations start from current data rather than last quarter's headlines. Subscribe to the newsletter to stay ahead of the next pricing move.