Summary

• Anthropic filed a Form S-1 on 26 October for a proposed Nasdaq listing under ANT; the share count, price and final valuation remain undetermined.
• First-half revenue reached $16.286 billion, against $8.945 billion of compute and infrastructure expense and a $693 million operating loss.
• Its $80.748 billion net loss was dominated by fair-value adjustments to financing instruments, which must be distinguished from operational spending.

Artificial intelligence developer Anthropic has publicly filed for a US stock market listing, revealing $16.3 billion in revenue during the first half of 2026 alongside substantial computing expenditure and financing arrangements that complicate the company’s reported financial results.

The developer of Claude submitted its Form S-1 registration statement to the US Securities and Exchange Commission on 26 October, seeking to list Class A shares on the Nasdaq Global Select Market under the symbol ANT.

The preliminary prospectus does not establish the number of shares to be offered or their price. The registration statement has not become effective, meaning a final valuation and completed flotation cannot be inferred from the filing.

Anthropic generated $16.286 billion in revenue during the six months ended 30 June 2026, compared with $1.163 billion in the equivalent period of 2025. Revenue for the whole of 2025 was $4.585 billion, illustrating the pace at which the company’s commercial business has expanded.

However, its filing also reports $8.945 billion of compute and infrastructure expenditure during the latest six-month period, alongside an operating loss of $693 million and a much larger net loss of $80.748 billion.

Computing costs remain substantial

Anthropic generates revenue from access to its AI models and related services, including subscriptions and usage-based arrangements used by developers and businesses.

Revenue growth depends on attracting customers, expanding their use of the company’s products and maintaining prices sufficient to cover the costs of developing and delivering increasingly capable models.

Those costs include computing resources used to train models, infrastructure supporting inference, specialist research staff and the operational expenses associated with delivering services to customers.

Compute and infrastructure expenditure rose from $2.334 billion in the first half of 2025 to $8.945 billion during the same period in 2026.

As a proportion of revenue, however, the figure declined from approximately 201% to 55%. That represents a substantial improvement in the relationship between reported computing expenditure and sales.

The percentage should not be interpreted as a universal measure of profitability for individual models. Different products may involve different infrastructure requirements and pricing arrangements, while expenditure on model development may support capabilities that generate revenue over an extended period.

The prospectus also shows an operating loss of $693 million during the first half of 2026, compared with $2.135 billion a year earlier.

Those results indicate improving operating economics, although they do not establish that profitability can be maintained as the company expands infrastructure commitments and invests in new models.

Operating performance remains sensitive to customer demand, the cost of acquiring computing resources and the proportion of available capacity generating revenue.

The headline net loss needs explanation

Anthropic’s reported net loss of $80.748 billion is considerably larger than its operating loss, principally because of accounting adjustments associated with financing instruments issued while the company remained privately held.

The prospectus records a $72.388 billion expense from changes in the fair value of convertible notes, derivative liabilities and simple agreements for future equity, commonly known as SAFEs.

According to the filing, the increase was driven largely by the rising fair value of outstanding convertible notes as the company’s estimated enterprise valuation increased.

Under the relevant accounting treatment, those changes are recognised in financial results even though they do not represent equivalent cash payments made during the reporting period.

Other financial instrument adjustments also contributed substantially to the difference between the company’s operating and net results. The prospectus includes an additional $8.757 billion of net other expense, principally associated with financing-related adjustments.

These charges cannot appropriately be described as money spent on computer equipment, research or ordinary operations.

They are nevertheless relevant to investors because the underlying instruments can affect the company’s ownership structure and financial obligations as it approaches public listing.

Some instruments may convert into shares or otherwise change their accounting treatment when a qualifying offering takes place, depending on the contractual terms.

The distinction between cash expenditure and valuation-related charges is therefore essential when evaluating the filing.

Anthropic also reported positive operating cash flow of approximately $3.9 billion during the first six months of 2026, a figure that differs from both its operating loss and its net loss.

Cash flow does not establish sustainable profitability on its own, particularly where changes in working capital or customer payment timing influence the result.

Infrastructure commitments extend beyond current costs

Developing advanced AI models requires access to specialised computing infrastructure, often through arrangements involving major cloud providers and hardware suppliers.

Anthropic has entered substantial agreements associated with computing capacity, including relationships with Amazon, Google and Microsoft.

Such arrangements can give a model developer confidence that resources will be available as customer demand grows, but they also create exposure if actual utilisation falls below expectations or contract costs become difficult to support.

Infrastructure agreements may include minimum purchasing commitments, leases and obligations that depend on future capacity becoming available.

Those different categories should not be combined into a single figure presented as an immediately payable debt. The timing and conditions of each commitment affect its financial significance.

The company’s need for infrastructure reflects the difference between developing an AI model and delivering it continuously to customers. A successful model may require further computing capacity as more organisations use it, even where the cost of producing an individual response becomes more efficient.

New model generations can also require additional training expenditure and technical investment before their commercial demand is established.

For investors, the economic assessment involves comparing those commitments with future revenue, the efficiency of model operation and the ability to generate sufficient cash from customer activity.

European businesses face related dependencies

Anthropic is an American company seeking a US listing, but its financial results have consequences for European organisations purchasing, supplying or competing in AI technology markets.

European software developers increasingly incorporate externally supplied models into business applications. Their operating costs can depend on pricing and technical arrangements established by the model provider rather than equipment they control directly.

If a provider changes its models, pricing or contractual terms, customers may need to alter applications or reassess whether particular workloads remain economical.

Applications relying on proprietary capabilities can also be difficult to transfer between suppliers without testing how different models behave.

The infrastructure requirements of large model developers have another European effect. Demand for computing capacity can support investment in data centres, networking and electrical infrastructure while increasing competition for suitable sites and specialist equipment.

Those opportunities also carry exposure to the commercial fortunes of a relatively small group of major customers.

A computing facility designed around the requirements of a particular AI developer may need extensive adaptation if demand changes or a customer alters its infrastructure strategy.

European policymakers concerned with technological sovereignty also face decisions about which parts of the AI supply chain are commercially realistic to develop domestically.

The scale of Anthropic’s expenditure illustrates the financial demands associated with frontier-model development, although it does not imply that every AI business needs a comparable investment.

Companies building specialist models, industrial applications and software tools can operate under different commercial arrangements, including the use of models supplied by larger developers.

Public investors face unusual governance arrangements

Anthropic’s proposed listing includes a governance structure intended to retain substantial founder influence while preserving a role for the company’s Long-Term Benefit Trust.

The preliminary prospectus describes a Founder LLC holding a special class of shares carrying 50.1% of voting power on matters submitted to stockholders generally.

The Long-Term Benefit Trust has defined rights relating to the appointment and removal of directors, creating an arrangement that differs from ordinary shareholder voting structures.

Those provisions reflect the company’s existing public benefit corporation status and its stated approach to governing advanced AI development.

However, a governance structure designed to address long-term safety considerations must coexist with the financial responsibilities of a publicly listed company seeking substantial investment.

Investors will need to assess how decision-making authority is allocated, which matters remain under founder control and how the trust’s responsibilities affect corporate oversight.

The prospectus also identifies risks associated with competition, infrastructure dependence, AI misuse, regulation and the possibility that the company may not achieve or sustain profitability.

Those disclosures are part of the framework through which potential investors can evaluate the business, rather than evidence that every identified risk will occur.

The proposed share price, offering size and final valuation remain undetermined. Anthropic’s filing establishes a more detailed financial record, but public market pricing will ultimately test investors’ confidence in its revenue growth, infrastructure economics and ability to operate profitably over time.