Chart – Anthropic's IPO: A Deep Dive into the Rapid Growth and Rising Costs of this Leading AI Company

Add Chart

Deborah Mary Sophia

Reuters, October 26 - Anthropic on Monday unveiled its initial public offering (IPO) filing in the United States, paving the way for a highly anticipated market debut. At a time when concerns are growing over the company's heavily debt-driven spending, the listing will test whether investors are willing to bankroll the artificial intelligence giant's expenses.

The company had submitted a confidential filing in June (link). In recent weeks, exclusive reports from Reuters have highlighted its grand AI ambitions (link), soaring costs (link), and concerns over "existential risks to humanity (link)" that could arise from advanced AI.

Below are several details about Anthropic’s financial situation, business structure, and its position in the global artificial intelligence race:

Background and Financial Situation

Anthropic was founded in 2021 by former OpenAI researchers, including siblings Dario and Daniela Amodei, and is the startup behind the "Claude" series of AI models. Claude encompasses four levels—Fable, Opus, Sonnet, and Haiku—powering Anthropic's chatbots, code assistants, and enterprise-focused Claude Cowork AI agents, among other products.

Anthropic initially focused on training models to accomplish programming tasks, propelling the company to the forefront of the AI race and helping attract a steady stream of enterprise clients.

Most of its revenue comes from enterprise clients who pay per use of the models, billed by the number of processed data “tokens”—a model Anthropic refers to as "usage-based billing." The company also generates revenue from paid subscription packages for customers, such as "Pro" and "Max."

This has resulted in Anthropic's revenue growing from $386 million a year earlier and $41 million in 2023, reaching $4.59 billion in 2025. For the six months ending June 2026, Anthropic’s revenue reached $16.29 billion.

But rapid growth has been accompanied by greater losses. Operating losses for the first half of 2026 were $693 million. In 2025, this figure expanded from $2.98 billion a year earlier to $8.06 billion.

https://www.reuters.com/graphics/ANTHROPIC-IPO/lgvdbeookvo/chart.png

Costs, Computing Demands, and Circular AI Investment

Computing capability is Anthropic's largest expense.

The company's costs include payment for compute fees to cloud service providers such as Amazon Web Services (AWS) and Google Cloud, expenses related to data center facilities and lease of data center capacity, equipment lease fees, and energy costs.

Over the past three years, Anthropic's spending on compute and infrastructure has far exceeded its revenue.

https://www.reuters.com/graphics/ANTHROPIC-IPO/zdpxeglrbvx/chart.png

The company has committed to pay over $500 billion to six partners over ten years (link) — making it one of the largest AI infrastructure buildout commitments ever. Anthropic states that many aspects of these agreements are non-cancellable, or must be paid regardless of actual usage.

The company is committed to paying Amazon AMZN.O at least $110 billion, Google GOOGL.O about $114.5 billion, and Microsoft $44.9 billion for cloud computing services over the coming seven to ten years.

https://www.reuters.com/graphics/ANTHROPIC-IPO/zjvqlmxwopx/chart.png

However, some of these companies are also its largest backers. Both Amazon and Google are among Anthropic's major shareholders, once again underscoring the prevalence of circular financing structures in the flow of capital within the artificial intelligence sector.

Amazon has invested at least $13 billion in Anthropic and agreed to reinvest up to an additional $20 billion. Google has pledged up to $40 billion in investments to Anthropic.

https://www.reuters.com/graphics/ANTHROPIC-IPO/zdvxeglqwpx/chart.png

Governance and Structure

Anthropic is a Public Benefit Corporation (PBC), meaning it is legally obligated to both pursue investor profit while developing artificial intelligence for the benefit of humanity. Most of its board members are elected by an independent entity.

The company is jointly led by CEO Dario Amodei and his sister Daniela—who serves as both president and board chair—and will continue operating as a PBC after listing.

The company is expected to remain under the control of its founders and an independent trust, an arrangement that may limit investor oversight rights and the influence of public shareholders. Anthropic stated that, in order to reduce the risks of artificial intelligence, control and capital should be concentrated in the hands of a few.

Under this arrangement, the majority of the seven co-founders will decide, through a single class F share (representing 50.1% of total voting rights), on key company matters, including the election of certain directors and other matters submitted to investors.

Competition

Anthropic’s most direct competitor is OpenAI, the developer of ChatGPT, whose launch in 2022 set off a global race for adoption and advancement of artificial intelligence technologies.

After Anthropic led the way in the enterprise market, OpenAI has further shifted its focus to products targeting developers and enterprise clients.

https://www.reuters.com/graphics/ANTHROPIC-IPO/dwvkmnkdrvm/chart.png

Anthropic is also competing against Elon Musk’s SpaceX, Alphabet’s Google, and Meta META.O. All these companies are investing billions of dollars in model development, data centers, and custom AI chips in the battle for generative AI supremacy.

Ramp data (reflecting a portion of the enterprise client's AI spending) shows that, in the week ending October 11, Anthropic accounted for 60.1% of AI token expenditures. As competitors ramp up investments in AI services, this startup has taken a leading position in the race.

https://www.reuters.com/graphics/ANTHROPIC-IPO/akpelroadvr/chart.png

(To facilitate non-native English speakers, Reuters has automated its reporting into several other languages. As automated translation may contain errors or lack required context, Reuters does not guarantee the accuracy of the automated translation text; it is provided solely for reader convenience. Reuters accepts no liability for any damages or losses arising from the use of the automated translation feature.)