Anthropic's reported $11.6B quarter puts OpenAI's 18% on the defensive before either lists

Anthropic's second-quarter revenue reportedly more than doubled to $11.6 billion while OpenAI's rose 18% to $6.7 billion, according to a Wall Street Journal report on August 18 ^[1]. OpenAI itself said quarter-on-quarter sales growth cooled to 18% while losses deepened ^[5], and that matters because investment bankers expect the first major AI IPO to set the precedent every other AI valuation gets measured against, with Anthropic's listing possibly arriving first ^[4].

Start with the arithmetic, because the headline numbers do not reconcile cleanly. On the reported quarterly figures, Anthropic's revenue is roughly 73% larger than OpenAI's ^[11]. On run rates, OpenAI has gone from about $20 billion at the end of 2025 to more than $40 billion, while Anthropic moved from around $9 billion to over $65 billion in late July ^[3], a jump of about 7.2 times ^[13]. Bloomberg put Anthropic's annualized figure above $65 billion, more than sevenfold higher than the end of last year and more than 50% above OpenAI's ^[2]; the run-rate spread is closer to 63% ^[12]. But four times an $11.6 billion quarter is $46.4 billion, about $19 billion short of the $65 billion run rate ^[24], which implies monthly revenue near $5.4 billion in late July against a second-quarter average of roughly $3.9 billion ^[25]. Cryptopolitan notes the obvious caveat: run rate is not audited annual revenue, and neither company files like a public one ^[10].

The valuations tell a different story from the growth rates. Anthropic's last round valued it at $965 billion against OpenAI's $852 billion ^[8]. Against reported run rates, that is about 14.8 times revenue for Anthropic and 21.3 times for OpenAI ^[14]. The talked-about targets, roughly $2 trillion for Anthropic and up to $1 trillion for OpenAI ^[9], work out at about 31 times and 25 times ^[15]. The slower-growing company is the more expensive one.

Adoption data cuts the same way, thinly. Ramp's AI Index for August 12 had Anthropic subscribed by 43.5% of US companies, up 1.1 points, versus OpenAI's 39.7%, up 0.23, and xAI at 4% ^[16]: a lead of 3.8 points ^[26]. The model-level data is less kind. Ramp economist Ara Kharazian called Anthropic's Fable 5 the best model to reach the market, yet it took only 6% of Anthropic tokens bought by businesses in its first month and 11.4% of spend ^[17], implying buyers paid roughly 1.9 times Anthropic's average per-token price to use it ^[27]. Fable 5 runs around $10 per million tokens, twice the competition ^[19]. OpenAI's GPT-5.6 Sol took 25% of tokens and 23% of spend ^[18], slightly cheaper than its own average mix ^[28]. Ramp cautions that its model-usage sample skews more technology-heavy than the broader index ^[20].

Then there is who is actually paying. In July the top 1% of firms spent an average of $7,400 per employee on AI, the top 10% spent $650, and the median company spent $11.95 ^[21]: the top percentile outspends the median by about 619 times ^[29], the top decile by about 54 times ^[30]. Kharazian's view is that further growth depends on those existing heavy spenders, who are also testing open-source and cheaper options and may force frontier labs to justify their prices ^[22]. Both firms are meanwhile in a price war with Chinese rivals selling marginally less advanced models at a steep discount ^[6]. "The US labs have cut the middle and are defending the top," an expert told the Financial Times ^[7]. Cryptopolitan reports Goldman Sachs' James Covello, head of Global Equity Research, raised similar concerns on May 11 ^[23].

Watch whether Anthropic's next reported quarter closes the gap between $46.4 billion annualized and a $65 billion run rate ^[24], whether Fable 5's token share rises at $10 per million ^[19], and whether median per-employee spend moves off $11.95 ^[21].