Anthropic will begin marketing its initial public offering in mid-October and complete the listing shortly before the US mid-term elections in November, according to sources familiar with the matter — a departure from the company's original timeline.

The delay poses a wider question for public investors: are they willing to pay the same premium prices for frontier AI that private backers have paid? Some investors estimate the company could be valued at close to $2 trillion, which would make Anthropic's listing one of the largest IPOs in history and the first major test of whether trillion-dollar private AI valuations can hold up in public markets.

A late-September prospectus and a $15 billion backstop

The most visible change is procedural. According to two sources who spoke to Reuters, Anthropic was expected to file its IPO prospectus as early as this week, but the documents likely will not arrive until later in September. That filing would mark the start of the final phase of the offering.

The company already has its legal clearance. On June 1, it said it had confidentially submitted a draft S-1 registration statement to the SEC, allowing it to move forward with the deal at any time once the regulator finishes its review, subject to market conditions and other factors. Confidential draft filings, permitted by the SEC for more than a decade, let companies work through regulator comments away from public view and give them flexibility on timing — which is precisely the lever Anthropic is now using.

Ahead of analyst presentations, Anthropic is also working to close a $15 billion revolving credit facility, Reuters reported. Facilities of this kind are commonly arranged ahead of large listings to provide liquidity and demonstrate financial flexibility to prospective investors. Morgan Stanley, Goldman Sachs, JPMorgan and Citi are among the banks working on the IPO. Anthropic and the banks declined to comment, and the sources cautioned that the schedule could still change.

Why the valuation question is live

Anthropic's most recent private financing valued the company at roughly $965 billion, ahead of OpenAI's $852 billion valuation. Some Anthropic backers are reportedly weighing a future worth near $2 trillion against an OpenAI target of up to $1 trillion, Cryptopolitan has reported.

That gap tracks a sharp divergence in reported growth. Citing the Wall Street Journal, Cryptopolitan reported that Anthropic's second-quarter revenue more than doubled to $11.6 billion, while OpenAI's rose 18% to $6.7 billion. Anthropic's annualized revenue run rate topped $65 billion by the end of July, versus more than $40 billion reported for OpenAI.

Enterprise demand points the same way. Ramp's August AI Index showed that in July, 43.5% of US companies purchased Anthropic subscriptions or tokens, compared with 39.7% for OpenAI. The mid-October pitch, in essence, is a claim that a faster growth trajectory and a modest edge in the enterprise segment can justify a valuation well above Anthropic's latest private mark.

The cost side that could cap the multiple

The pressing question is how much Anthropic must spend to sustain that growth. Ramp found that in its first month, Fable 5 accounted for 6% of the Anthropic tokens bought by companies and 11.4% of model expenditure, despite carrying the highest price of any model the company offers.

The pricing gap is significant. Anthropic's official Fable 5 pricing is $10 per million input tokens and $50 per million output tokens. OpenAI lists its GPT-5.6 Sol at $4 and $20 respectively, making Fable 2.5 times more expensive. Better performance does not automatically mean buyers must pay more for it.

Sector-wide economics are getting harder. Gartner's "Inference Paradox" predicts that although unit token economics will improve, cost per agentic workflow will rise more than fivefold through 2028. Goldman Sachs analyst James Covello has described the overall spending trend as "unprecedented and unsustainable."

At the same time, OECD research shows quality-adjusted model prices are falling, and McKinsey's latest State of AI survey suggests company-wide AI deployment still lags the rapid adoption shown in headline figures. Until Anthropic publishes its S-1 with audited financials, Morningstar notes, investors have limited visibility into key measures such as gross margins, cash flow and customer concentration.

A build-out priced mostly in private

Staying private longer carries its own risks. According to Forge's private market study, Anthropic, OpenAI and xAI reached $100 billion valuations in an average of five years, while the previous generation of companies took roughly 16 years. Much of the value investors once expected to capture after an IPO has therefore already been created while the firms remained private. Going public may increasingly serve as an exit opportunity for early investors and employees rather than the point at which value creation begins.

Even with uncertain future returns, the overall market continues to expand rapidly. Gartner forecasts global spending on AI models and platforms will grow 63.4% to $64.3 billion by 2026. Goldman Sachs Research says global AI investment will exceed $1 trillion this year, and PwC's Global Data Centre Outlook anticipates $31.6 trillion in spending through 2050.

Anthropic's IPO will show whether public investors are prepared to buy in at valuations set in private markets. SpaceX's June IPO, worth $1.77 trillion, offers the closest comparison — and if Anthropic's offering approaches $2 trillion, it will inevitably be measured against SpaceX's result. The S-1 filing expected later in September, with its audited financials, will give investors their first hard look at the numbers behind the private valuations.