Anthropic, the artificial intelligence company behind Claude, has made its initial public offering filing public, handing Wall Street concrete numbers to weigh ahead of a potential Nasdaq debut under the ticker ANT.

The company submitted its registration paperwork to the U.S. Securities and Exchange Commission privately in June, giving regulators time to review the documents before the rest of the market gained access. The public filing now lays out the financial picture behind one of the AI industry's most closely watched businesses. A public registration statement opens detailed financials, risk factors, and share-class terms to any prospective investor ahead of pricing — the disclosure step that customarily precedes a roadshow and listing.

Anthropic opens its books as Claude brings in billions

Dario Amodei founded Anthropic in 2021 with a group of researchers who left OpenAI, now one of its biggest competitors. Amodei runs the company chief executive, while fellow co-founder Daniela Amodei serves as president and board chair.

Businesses have been spending heavily on Claude, betting it can help employees write software, move through routine tasks faster, and build services they could not easily offer before. Claude Code has been a major driver of growth in 2026, helped by new models including Claude Fable 5 and Mythos 5.

Selling powerful AI, however, is hardly a quiet business. Anthropic is fighting Google and OpenAI for customers, while its positions on safety, government rules, and national security have drawn criticism. Those disagreements have also produced two high-profile clashes with the U.S. government in recent months.

The financial figures explain why investors are examining the deal closely. Anthropic posted revenue of $16.3 billion in the first half of 2026, compared with $1.2 billion in the same period the previous year. That is a more than thirteenfold year-over-year increase. Operating losses narrowed to $693 million, a $2.1 billion improvement from the year before.

One further earnings figure requires clarification. Adjusted operating income stood at $1.6 billion, a non-GAAP metric that reflects adjustments made on top of the standard result. Under conventional accounting rules, the company still reported operating losses.

Those margins cannot be considered stable. Costs can rise suddenly when the company buys access to additional computing resources before fully utilizing them, and AI development carries substantial infrastructure expenses.

Most customers pay based on how much AI they consume. That business brought in $13.1 billion, or 80% of first-half revenue, while subscription plans delivered another $3.2 billion, accounting for the remaining 20%. Usage prices vary with the model chosen and the number of tokens, the units used to measure what the AI processes.

Anthropic operates direct sales channels but also reaches buyers through Amazon Web Services, Google Cloud, and Microsoft Azure. Those cloud platforms accounted for 48% of Anthropic's revenue in the first six months. The platforms invoice customers, collect payments, and then remit Anthropic its share minus their fees. The overlap puts Google on both sides of Anthropic's business — a rival for customers and, through Google Cloud, one of the channels behind that 48%.

The company also reported $10.8 billion in remaining performance obligations as of June 30 — in simple terms, revenue anticipated from contracts with non-cancelable commitments. Anthropic expects to realize about 75% of that total over the next 12 months, based on customer utilization and its own delivery.

Wall Street gets a complicated vote and an expensive AI bet

Buying shares will not give investors much say over the company. Anthropic plans to remain a Delaware public benefit corporation, meaning it must weigh its stated public mission alongside the financial interests of shareholders. Tiered share structures of this kind have become a familiar feature of founder-controlled technology listings.

Its Long-Term Benefit Trust generally controls who occupies a majority of board seats through Class T shares. A separate company representing the founders is expected to hold a Class F share carrying 50.1% of power on general shareholder decisions, including elections for directors outside the trust-controlled seats. Ordinary investors will receive Class A shares with one vote per share — a great deal of voting structure to absorb before hitting the buy button.

The timing is tricky as well. Anthropic's potential debut follows SpaceX's $85.7 billion IPO, the biggest ever, by just months. Shares of Elon Musk's company have swung sharply since listing on Nasdaq and ended last week slightly below their June opening price. That trajectory gives investors the nearest recent reference for how the market has absorbed a mega-cap technology debut.

Speculation is mounting over whether demand for AI products justifies the trillions of dollars being poured into infrastructure. Firms that rent out computing power have felt the pinch. CoreWeave and Nebius, known as neocloud firms, each lost more than 30% of their market valuation in October amid heightened focus on borrowing risks in the data center industry.

Still, Monday's pre-market session showed signs of strength as crude oil prices moved lower. Futures for the Dow Jones Industrial Average rose 209 points, or 0.4%; S&P 500 futures gained 0.4%; and Nasdaq-100 futures climbed 0.7%. Among chip stocks, Nvidia was up 1.4%, Micron rose 2.3%, Broadcom increased 1.5%, and AMD climbed 1.3%.