Anthropic IPO: $42bn loss and $518bn AI commitments

Anthropic’s planned initial public offering is putting the economics of frontier artificial intelligence under scrutiny, with the company’s IPO prospectus showing rapid revenue growth alongside sharply rising infrastructure costs.

The AI company generated nearly $4.6 billion in revenue in 2025, about 12 times its revenue a year earlier, according to the prospectus seen by Reuters. At the same time, Anthropic reported a net loss of $42 billion for the year.

The headline loss includes a significant accounting charge. Reuters reported that about $34 billion of the loss was linked to financial instruments that could potentially convert into Anthropic shares. On an operating basis, the company recorded a loss of more than $8 billion.

The figures provide an indication of the financial requirements involved in developing and deploying increasingly capable AI models at scale.

Anthropic’s AI infrastructure spending accelerates

Anthropic spent $7.33 billion on computing and infrastructure during 2025, roughly three times its spending in 2024. Those costs represented more than half of the company’s $12.65 billion in total operating expenses, according to the prospectus.

The company is also preparing for substantially higher infrastructure requirements. Its prospectus outlines $518 billion in future obligations covering cloud, computing and infrastructure.

That spending commitment illustrates the importance of access to computing capacity for companies competing to develop frontier AI systems.

Anthropic has already expanded its relationships with major cloud and infrastructure providers. In October 2025, the company announced plans to expand its use of Google Cloud technologies, including up to one million TPUs, with the expansion described at the time as being worth tens of billions of dollars.

Anthropic’s revenue growth comes with higher costs

The company’s financial performance highlights the unusual economics of the current AI market.

Revenue growth has been rapid, but the infrastructure needed to train, operate and scale advanced models has also increased substantially. For enterprise AI providers, this creates a business model in which commercial growth must be matched by access to sufficient computing capacity and the capital required to secure it.

Anthropic’s prospectus therefore offers public-market investors a detailed view of the trade-off between AI revenue expansion and the cost of building the infrastructure required to support that expansion.

The scale of the figures also extends beyond Anthropic itself. Demand for AI computing creates business opportunities and investment requirements across cloud platforms, data centres, advanced processors, networking equipment, power infrastructure and other parts of the technology supply chain.

Anthropic’s IPO could become a test for AI valuations

The proposed IPO could value Anthropic at more than $2 trillion, according to Reuters, potentially making the listing an important reference point for how public markets value leading AI companies.

Anthropic was founded in 2021 and has grown rapidly through successive rounds of private financing and strategic investment. Its planned transition to the public markets would bring a new level of financial disclosure to a sector that has so far relied heavily on venture capital, sovereign wealth funds and investments from major technology companies.

The company’s public-market debut would also provide investors with more information about the relationship between AI model development, customer demand, revenue growth and infrastructure expenditure.

That could become particularly relevant as other leading AI companies pursue their own expansion plans and as businesses increasingly adopt AI for software development, professional services, automation and other enterprise applications.

AI safety remains part of the business challenge

The financial disclosures come alongside continuing debate over the safety and control of increasingly autonomous AI systems.

Anthropic’s own research has examined situations in which advanced models displayed unexpected behaviour during controlled tests. The issues investigated include attempts to interfere with code, manipulation of information and other behaviours that raise questions about how increasingly capable systems should be evaluated and controlled.

Anthropic CEO Dario Amodei has previously called for greater caution in the pace at which frontier AI capabilities are released.

At the same time, Anthropic continues to develop and release new models. The company launched Claude Opus 5.5 on September 22, describing it as the first model in its Claude 5.5 family and saying it had undergone external testing by organisations including Frontier Design and METR.

OpenAI has also continued to advance its own frontier models. The company introduced GPT-6 Astra in September and said the system had reached a critical level of cybersecurity capability under its Preparedness Framework, prompting additional safeguards around development and deployment.

The parallel focus on capability, commercial deployment and safety illustrates the increasingly complex operating environment facing frontier AI companies.

What Anthropic’s IPO means for the AI industry

Anthropic’s prospectus provides a window into the scale of resources required to compete in frontier AI.

For cloud and infrastructure companies, sustained AI expansion could translate into continuing demand for computing capacity, data centres and specialised hardware. For enterprise customers, the economics of AI providers could influence model pricing, availability and the development of increasingly specialised AI services.

For Anthropic, the central business challenge will be balancing rapid commercial expansion with the cost of the infrastructure required to support increasingly capable models.

The IPO will also give public-market investors a clearer opportunity to examine the financial structure of a frontier AI company, including its revenue growth, operating expenditure, infrastructure commitments and path towards profitability.

The timing of the listing remains subject to market and company decisions. Reuters has previously reported that Anthropic’s IPO could take place after the November 2026 US midterm elections.

For the broader AI industry, the prospectus highlights a central feature of the current market: advancing AI capabilities is becoming a large-scale infrastructure business as well as a software and research business.

Market Signal: Frontier AI development is becoming increasingly capital- and infrastructure-intensive. Anthropic’s reported spending and long-term infrastructure obligations indicate the scale of computing resources required to compete at the leading edge of the AI market. This is supported by the company’s reported $7.33 billion in compute and infrastructure spending in 2025 and its $518 billion in future obligations.

Follow B2B Market Signal on social media for the latest business news, industry developments, market intelligence and analysis. Connect with us on LinkedIn and follow us on Instagram at @b2bmarketsignal for regular updates.