Artificial intelligence firm Anthropic has officially unveiled its long-awaited listing prospectus, setting the stage for a blockbuster initial public offering (IPO) on the Nasdaq under the ticker symbol ANT.

While the IPO serves as a major referendum on public market appetite for visionary AI ventures, the company’s 240,000-word S-1 filing reveals an extraordinary financial profile, massive infrastructure bets and a complex corporate structure.

Here are five essential things prospective investors need to know before buying into Anthropic.

1. Soaring revenues backed by rapid efficiency gains

Anthropic’s financial disclosures paint a picture of staggering growth. Revenue surged from $4.6bn last year to more than $33bn in the nine months ending September. However, growth has begun to taper off, moving from a doubling between the first two quarters of 2026 to roughly 50% growth between Q2 and Q3.

Critically, the company is aggressively pushing toward profitability by optimising its compute utilisation, leveraging its Claude AI models to train newer iterations. This efficiency drive has paid off: while Anthropic spent $2.50 on computing power for every $1 of revenue in early 2025, that figure dropped to under 50 cents by Q2 2026, yielding an operating income of $1bn on an 8% operating margin.

Despite these gains, the company warns that customer concentration remains a risk, with just five customers accounting for 25% of revenue in the first nine months of the year.

2. Astronomical infrastructure commitments worth $800bn

Computing power is identified as the “critical input” for Anthropic’s operations, driving a wave of unprecedented data centre and hardware agreements stretching out to 2045. The firm has signed commitments exceeding $800bn, which the founders describe as “bold but sane.”

Major deals include a for custom TPU (Tensor Processing Units) processors co-developed with Google, up to $20bn with AMD, and contracts with cloud providers including Nscale and Lambda worth up to $86bn. Additionally, agreements with SpaceX total up to $120.5bn through 2029, while tech giants Google, Amazon and Microsoft are locked in to provide at least $269.4bn in cloud compute over the coming decade.

3. A ‘novel’ governance structure defies normal shareholder control

Anthropic’s corporate governance diverges sharply from standard public market norms, granting over major company decisions. At the apex of the structure sits a “Long-Term Benefit Trust”, whose trustees include former Federal Reserve Chair Ben Bernanke and Harvard physics professor Christopher Stubbs, which holds the power to elect and remove four out of seven board members.

Meanwhile, co-founders Dario and Daniela Amodei, alongside five other co-founders, retain day-to-day control via a special class of stock granting them 50.1% voting power. Prospective buyers are explicitly warned in the prospectus that founders’ decisions “may run counter to investors’ expectations” as profit maximisation is permanently balanced against the perceived “public good.”

4. Tech giants stand as massive early beneficiaries

Assuming Anthropic achieves a massive , its early tech backers are poised for extraordinary windfalls. Amazon is its largest hyperscaler backer, holding a stake worth approximately $400bn under fully converted notes (though capped at 12% initially unless waived).

Google holds a roughly 7% stake, with options to invest up to $20bn more. Financial backers including Spark Capital and Coatue also sit on multi-billion-dollar gains. Unlike traditional IPOs that lock in early investors for 180 days, Anthropic has structured its debut to allow non-founder directors, officers, and select employees to begin selling portions of their equity across five release windows within the first 76 days of trading.

5. Unprecedented disclosures on catastrophic and existential risks

In stark contrast to boilerplate legal disclaimers, outlines terrifying theoretical scenarios. The company warns that its advanced models may pose “catastrophic or existential risks to humanity,” potentially providing “meaningful assistance to malicious actors seeking to develop chemical, biological, or other dangerous weapons.”

The filing also cautions that Anthropic could eventually lose control of its own products, noting that models might exhibit “unexpected… self-preserving behaviours, including attempts to resist shutdown, conceal or manipulate information” or generate coercive, deceptive outputs resembling “blackmail.” The company openly acknowledges that these systemic risks may be impossible for the commercial insurance industry to underwrite.

For latest tech stories go to TechDigest.tv