11
Anthropic’s planned public offering is shaping up to be more than another technology IPO. It could become a defining test of how financial markets value artificial intelligence at a moment when the technology’s ambitions are expanding faster than its economics.
The company’s IPO prospectus presents an extraordinarily ambitious vision. Anthropic argues that artificial intelligence could transform the global economy on a scale comparable to, or even greater than, industrialisation, electricity and the internet. Yet the same document exposes the enormous financial requirements behind that ambition.
Anthropic’s revenue growth is striking. The company generated nearly $4.6 billion in revenue in 2025, roughly twelve times its 2024 figure. But rapid growth has come with extraordinary costs. Anthropic reported a net loss of about $42 billion last year, although roughly $34 billion of that figure was an accounting charge associated with the valuation of financing instruments rather than an operating cash expense. Its operating loss nevertheless exceeded $8 billion.
The more important issue may be what lies ahead. Anthropic disclosed plans involving approximately $518 billion in future cloud, computing and infrastructure obligations. In 2025 alone, it spent $7.33 billion on compute and infrastructure, accounting for more than half of its total operating expenses.
This illustrates the central economic dilemma of frontier AI. Building increasingly capable models requires enormous quantities of computing power, specialised chips, data-centre capacity and energy. The companies developing these systems therefore face a business model in which demand can grow rapidly while the cost of satisfying that demand grows alongside it.
That does not necessarily make Anthropic’s strategy unsustainable. Technology industries have repeatedly experienced periods in which infrastructure investment preceded mass-market returns. But AI is operating at an unusual scale, making capital efficiency an increasingly important question for investors and policymakers.
Anthropic’s prospective valuation adds another dimension to the debate. Its IPO could value the company at more than $2 trillion, according to the Reuters report. Such a valuation would place enormous expectations on future growth and profitability. The offering could therefore become an important reference point for the wider AI industry, particularly for competitors such as OpenAI.
There is another risk that deserves attention: customer concentration. Anthropic disclosed that nearly a quarter of its 2025 revenue came from just two customers, while many major customers were not committed to long-term contracts. That creates a potential vulnerability if corporate AI spending slows, customers switch providers or model capabilities become increasingly commoditised.
At the same time, Anthropic is confronting a question that cannot be measured simply through revenue or market capitalisation: AI safety. The company’s own research has identified instances in controlled testing in which increasingly autonomous systems displayed unexpected or potentially harmful behaviour. Anthropic has also warned investors that more advanced AI could create serious risks.
This produces a striking tension. The company is simultaneously warning about the risks associated with increasingly powerful AI and competing aggressively to develop and commercialise more capable models. That tension is not unique to Anthropic; it is becoming a defining issue for the entire industry.
CEO Dario Amodei has argued that the development and release of increasingly powerful AI capabilities should proceed more cautiously. Yet commercial competition places pressure on companies to release new systems quickly, particularly as OpenAI, Google, Meta and xAI pursue their own strategies.
The forthcoming IPO will therefore be about more than whether Anthropic can generate impressive revenue growth. It will ask investors to assess whether the economics of frontier AI can ultimately justify the extraordinary infrastructure expenditure required to build it.
If Anthropic succeeds, its public listing could help establish a new financial model for the AI economy, where enormous upfront infrastructure costs are supported by equally enormous demand for intelligent software and services. If growth fails to keep pace with spending, however, the prospectus may come to represent an early warning about the limits of the current AI investment boom.
The most important lesson is that AI’s future will not be determined by technological capability alone. It will also depend on economics: who pays for computing power, how efficiently it is used, how durable customer demand becomes, and whether AI companies can convert extraordinary technological progress into sustainable businesses.
