The AI boom has run for three and a half years without a single pure-play, public company to measure it by. That gap is about to close, and Deutsche Bank argues its closing will change the boom more than any model release has.
In a note published this week, research analyst Adrian Cox makes the case that the expected listings of OpenAI and Anthropic will reset the industry. Both companies filed confidentially for IPOs in early June, each reportedly aiming to raise around $60 billion at a valuation above $1 trillion. News last week that OpenAI is weighing a delay to its own sale, first expected later this year, shows how high the stakes have climbed.
Why the listings matter beyond the cash
Cox lays out three reasons the flotations would matter. Public reporting would force transparency on financials and business models that private disclosure has kept opaque. Access to equity and debt markets would broaden funding beyond the tangled, multi-billion-dollar deals struck between the leading firms and their backers. And listed status would raise accountability, dragging the companies into public scrutiny over jobs, copyright and safety.
The scale of the funding need explains the urgency. Deutsche Bank notes that capital spending by the hyperscalers looks set to rise by two-thirds to more than $750 billion this year. Even cash-rich Google turned to debt markets, raising almost $32 billion in under 24 hours in February, then $84.75 billion in the largest equity capital markets deal on record on June 1.
Anthropic pulls ahead on revenue
The competitive picture has shifted fast. Anthropic's revenue reached an annual recurring rate of $47 billion in May, overtaking OpenAI, which Cox puts at roughly $25 billion to $33 billion. Both figures dwarf where they sat at the end of 2024, when Anthropic ran at $1 billion a year and OpenAI at $4 billion.
Usage tells a similar story. The Claude website drew almost one billion visits in May, up from 200 million in January, and Claude Code now accounts for slightly more than half the business coding market. Anthropic raised $65 billion at a post-money valuation of $965 billion, leapfrogging OpenAI's $122 billion round struck at $852 billion in March.
The pressure from cheaper rivals
The threat to both is price. Chinese open-source models now match Western performance for a fraction of the cost. Deutsche Bank cites Z.ai's GLM 5.2, which scores 51 on Artificial Analysis's Intelligence Index at 48 cents per task, against $2.75 for the top Claude model and $1.03 for OpenAI's GPT-5.5.
Enterprise customers have started to balk at real-world costs. Deutsche Bank reports that Amazon, Walmart and Cisco have reined in token use, while Uber burned through its annual AI budget in four months. Both OpenAI and Anthropic are said to be weighing price cuts that would squeeze margins.
The lesson from Netscape and Amazon
Cox is careful not to declare a bubble. The Shiller CAPE ratio for the S&P 500 recently topped 40, nearing the 44 it hit at the dot-com peak. But the companies coming to market are older and more profitable than their 1999 counterparts, with a median age of 12 years last year against about 4.5 at the height of the dot-com frenzy.
History offers a warning and a consolation. Deutsche Bank recalls that Netscape's 1995 listing drew immediate cries of froth, yet the Nasdaq did not peak for almost five years. Amazon debuted at 7.5 cents a share in 1997, split-adjusted, and took a decade to reclaim its early high before reaching almost $275 last month.
The point stands. A public OpenAI or Anthropic would finally let investors bet on the horses rather than, as Cox puts it, the jockeys and the track. Whether they can absorb hundreds of billions in rapid-fire demand is the question no filing can answer in advance.
