OpenAI could face more hurdles for IPO than Anthropic
As artificial intelligence giants inch towards the public markets, copyright litigation is becoming an increasingly influential factor in how investors assess risk.
And with that in mind, Anthropic appears to be starting from a less complicated position than its larger rival, OpenAI.
Analysts at Bloomberg Intelligence have said the Claude founder faces “significantly less legal uncertainty” over copyright claims, a distinction that could shape the timing of any public listing, widely expected to be no later than 2026.
Both firms are burning capital at scale, and at dizzying rates, while depending on sustained investor confidence to fund the huge costs of computing and infrastructure.
But in both cases, their exposure to copyright laws is starting to diverge in ways that public markets won’t overlook.
Why it matters to investors
For IPO investors, the concern is whether the potential liability can be bounded.
Tamlin Bason, industry analyst at Bloomberg Intelligence, told City PM: “Unresolved copyright litigation is a central risk to AI IPOs. Not because the claims are guaranteed to succeed, but because the potential litigation is so hard to quantify”.
Under US copyright laws, statutory damages can reach up to a staggering $150,000 (£111,384) per infringed work.
Given that large language models (LLMs) are trained on millions of texts, images and audio clips, the potential exposure is huge, even if courts eventually accept fair use claims.
“Fair use is a plausible defence”, Bason added, “but it is by no means certain to work in every instance. And even where it does, it is slow to develop”.
That mismatch between legal timelines and IPO calendars is what concerns underwriters.
Anthropic’s narrower exposure
Anthropic’s remaining disputes are relatively contained and largely limited to negotiations with music publishers and Reddit.
“If Anthropic manages to settle its remaining disputes”, Bason said, “it would not have active litigation posing the sort of massive liability risk that we have seen in this area”.
Such settlements would not eliminate future risk, but they would convert exposure into a defined cost.
“That would effectively clear its docket”, Bason added, “leaving it with a defined cost structure, rather than existential risk”.
Anthropic has already taken visible steps towards being IPO-ready, like hiring Wilson Sonsini, the law firm that advised public listings for giants like Google and LinkedIn.
The company also hired a chief financial officer with experience guiding Airbnb through its listing back in 2020.
OpenAI: A longer road ahead
On the other hand, OpenAI is facing a broad set of claims from authors, publishers and news corporations, none of which has yet delivered a ruling that definitively affirms its right to use copyrighted material for training without permission.
“OpenAI currently lacks any court ruling affirming its right to use copyrighted content for training without permission”, Bason said.
While OpenAI has struck licensing deals with major content owners, such as Disney, that activity might complicate its legal stance.
“If anything, its licensing activity may suggest an implicit recognition that rights holders deserve compensation”, Bason added.
The most consequential case is the multi-district litigation overseen by Judge Sidney Stein in New York. While briefing on summary judgment is not expected to conclude until October 2026, with trials potentially stretching into 2027.
“Given the complexities of the litigation”, Bason said, “we would expect the legal timeline to lengthen rather than shorten”.
That could prove difficult for an IPO process that relies on clarity.
Precedents, but no easy parallels
Public markets have, in the past, priced in unresolved legal risk.
Uber and Lyft both floated while questions over driver classification remained unsettled.
But copyright law operates on different dynamics, particularly when potential damages scale with volume.
“The worst-case scenario”, Bason warned, “would be courts finding that it was not fair use to use copyrighted content to train LLMs”.
So far, judges have accepted fair use claims in narrow circumstances, including in cases involving Anthropic and Meta.
But the Supreme Court has emphasised that fair use is highly fact-sensitive, leaving open the possibility of conflicting outcomes.
None of this guarantees Anthropic a smoother listing, nor does it rule out OpenAI going public first.
Both continue to grow rapidly, while facing challenges in forecasting long-term margins in an increasingly competitive market. However, the difference may come down to which firm can remove enough legal and financial uncertainty for public investors.