Anthropic’s $2tn valuation question: what would the AI firm need to earn to justify an IPO?

There is a useful moment in every technology boom when the conversation stops being about possibility and starts being about arithmetic.

Anthropic appears to be approaching that moment. According to the discussion source, some backers expect the privately held AI lab to go public in October with a targeted valuation of $2 trillion or higher. The company has filed confidentially with the US Securities and Exchange Commission, but it has not publicly set a timeline.

That matters because Anthropic is not just another software company. It is one of the frontier AI labs building general-purpose models, best known publicly for Claude. If a $2 trillion IPO target is even close to the mark, it would tell us a lot about what investors think the next phase of AI is worth.

It would also raise a much less glamorous question: how much profit does an AI company need to make before a valuation like that looks sensible?

Why Anthropic’s reported $2 trillion valuation is such a big ask

The reported numbers are striking. The discussion says Anthropic was valued at $965 billion when it reported a Series H funding round in May. A $2 trillion IPO valuation would more than double that figure.

It would also, according to the same source, eclipse SpaceX’s $1.77 trillion IPO in June. The Fortune report frames the central challenge plainly: a valuation at this level puts Anthropic in the same mental category as the largest and most strategically important technology businesses.

The awkward part is profitability. The source says Anthropic is not making money yet. That does not mean the business has no value. Plenty of major technology companies have spent heavily before becoming profitable. But it does mean investors are being asked to believe in a future earnings machine, not a current one.

The profit maths behind a $2 trillion AI IPO

The cleanest way to understand the issue is to look at earnings multiples. A price-to-earnings ratio, or P/E ratio, compares a company’s valuation with its annual profit. If a company is valued at 25 times earnings, investors are effectively paying £25 for every £1 of annual profit.

The discussion says the average Nasdaq 100 company trades at roughly 34 times trailing earnings and 25 times forward earnings. On that basis, a $2 trillion Anthropic would need annual profits in the region of $59 billion to $79 billion to keep pace with those multiples.

That is the crux of the debate. It is not whether AI is important. It is whether Anthropic can grow from a business that is not yet making money into one producing Amazon-style annual profits.

There are two ways investors can justify that gap. One is to believe Anthropic’s future revenue and margins will expand dramatically. The other is to believe that today’s normal valuation rules do not apply to frontier AI companies. The first argument is ambitious. The second is dangerous if used lazily.

What investors are really buying with frontier AI labs

When investors value a frontier AI lab, they are not only valuing today’s product subscriptions. They are also valuing model capability, enterprise demand, infrastructure partnerships, developer ecosystems, brand trust, and the chance that AI becomes embedded across a huge range of workflows.

That is why comparisons with ordinary software companies can feel incomplete. A frontier model company has the potential to sit underneath customer service, coding, research, data analysis, education, operations, marketing, and internal knowledge work. If it becomes a default layer in business technology, the revenue opportunity is large.

But the cost side is equally important. Frontier AI is capital intensive. Models require substantial computing infrastructure, specialist talent, ongoing research, safety work, and distribution. The source also says Bloomberg reported Anthropic is in talks to buy AI startup Decart AI for $6 billion, although the final status and full details are not disclosed.

That is the tension. Investors may be pricing in platform economics, but frontier AI labs still have to prove they can generate durable profits after infrastructure costs, competition, and pricing pressure.

Why OpenAI’s IPO timing matters to the Anthropic story

The discussion also says OpenAI filed confidentially after Anthropic, but is not expected to IPO until 2027. That creates an interesting race, although it is not purely about who lists first.

The more important contest is credibility. Public market investors will want to understand growth rates, margins, compute costs, enterprise adoption, product stickiness, and governance. They will also want a story about why one model provider can defend its position against rivals.

I explored that broader market dynamic in OpenAI vs Anthropic and the AI IPO race. The key point is that an IPO forces private AI narratives into public market discipline. Once a company is listed, quarterly numbers start to matter.

That could be healthy. Private valuations can drift into storytelling. Public markets, imperfect as they are, eventually ask for evidence.

What this means for UK businesses using AI

For UK business owners and technology leaders, the lesson is not to obsess over whether Anthropic is worth $2 trillion. The practical lesson is that the AI market is entering a more expensive, more competitive, and more financially accountable phase.

That has several implications.

• AI products may change quickly. If companies are under pressure to justify huge valuations, expect rapid product launches, pricing experiments, and enterprise packaging.
• Costs need watching. AI usage can look cheap at pilot stage and become material when rolled out across a team or customer base.
• Vendor risk matters. Do not build critical workflows with no exit plan. Keep prompts, data flows, and integrations documented.
• Data protection remains non-negotiable. UK organisations still need to think carefully about personal data, client confidentiality, and UK GDPR obligations when adopting AI tools.
• Productivity claims need evidence. Measure time saved, quality improved, and risk reduced. Do not rely on investor enthusiasm as proof of value.

The best AI adoption strategies are boring in the right ways. Start with a business problem, run a controlled pilot, measure the outcome, and decide whether to scale. A supplier’s valuation does not answer those questions for you.

The compute wars sit behind the valuation debate

One reason AI valuations have become so large is that the infrastructure race is enormous. Frontier AI companies need access to compute, data centres, chips, and strategic partners. That makes the sector look less like classic SaaS and more like a hybrid of software, infrastructure, and industrial strategy.

That is why deals and partnerships around AI compute matter. The strategic question is not simply who has the best chatbot today. It is who can afford to train, serve, and improve models at scale while maintaining reliability and margins.

For more context on that infrastructure angle, see my breakdown of Google’s reported Anthropic bet and the AI compute wars. The valuation debate makes more sense when you remember that model capability and infrastructure access are now deeply linked.

A useful reality check for the next phase of AI hype

Anthropic may become one of the defining companies of the AI era. It may also face the same brutal commercial test as every ambitious technology business: can it turn demand, capability, and investor confidence into sustainable profit?

The reported $2 trillion valuation is not impossible simply because it is large. But it does require a very large belief about future earnings. Based on the multiples cited in the discussion, the company would need tens of billions of dollars in annual profit to sit comfortably alongside large-cap tech peers.

That is the reality check. AI can be transformative and still be overpriced. A company can be strategically important and still have to prove its economics. For UK readers, the sensible position is neither cynicism nor blind belief.

Use the tools where they work. Track the companies with interest. But when the numbers get this big, always come back to the same plain question: where will the profits come from?