Anthropic projects $900B in AI infrastructure spending over two decades

The cost of the AI build-out

Anthropic is preparing for a period of intense capital expenditure as it seeks to build out the necessary infrastructure for advanced artificial intelligence. According to its investor prospectus, the company expects to spend $900 billion over the next two decades on chips, power, and other computing resources. This spending is necessary to develop increasingly sophisticated models like Claude.

The scale of this investment is among the most aggressive corporate bets in the current technology landscape. Goldman Sachs, the investment bank leading the company's initial public offering (IPO), has privately told investors that Anthropic could burn through more than $65 billion next year. This would be followed by another $41 billion in 2028 as the company continues its infrastructure build-out.

While the company has recently shown some financial stability, these upcoming costs represent a significant shift in its financial picture. Anthropic has already raised $141 billion privately from investors and has generated cash from its business in the first nine months of this year. However, the new spending targets indicate a move into a high-burn phase to maintain its competitive edge against rivals like OpenAI and Alphabet.

Revenue targets and growth

To offset these massive costs, Anthropic is aiming for a dramatic increase in its annual revenue. Goldman Sachs forecasts that the company could surpass $440 billion in annual revenue by 2029. This target would represent more sales than Apple generated in the previous year, marking a significant leap for a relatively young company.

Anthropic has already seen rapid growth, with quarterly sales jumping from about $1 billion to $17 billion in a single year. This growth was largely driven by heavy adoption from software developers rather than individual consumers. The company is now attempting to expand its reach to a wider range of white-collar employees to automate various work tasks.

The company's strategy involves moving beyond simple chatbots to tools that can participate in work meetings or manage spreadsheets. Executives believe these products will be essential for capturing enough market share to sustain the company. If successful, the revenue surge is expected to provide enough cash flow to cover the high costs of model development by 2029.

The model treadmill risk

Analysts have expressed concerns regarding the sustainability of Anthropic's current business model. Some observers have described the company's pace of releasing new versions of Claude as a "model treadmill." This suggests a constant need to innovate and release new products just to keep up with customer expectations and competition.

A significant risk identified by experts is the lack of long-term customer loyalty. The prospectus shows that the vast majority of Anthropic's revenue comes from models released in just the last six months. If customers do not stay committed to the platform, the company could face a situation similar to Netflix, where it must constantly spend on new content to retain subscribers.

Anthropic argues that its spending creates a reinforcing cycle that will eventually lead to profitability. The company claims that as models improve, the costs for both the company and its customers will decrease. They also use their own AI to automate the development of new models, which they say is speeding up their pace of discovery.

Public debut and hidden costs

Anthropic's upcoming listing on the Nasdaq could value the company at $1.4 trillion. This public debut would provide the company with an additional $100 billion to spend on its goals. It also makes it easier for the firm to raise even more capital to fund its long-term commitments to infrastructure and research.

However, the move to a public company brings new expenses that are not yet reflected in current accounting. The company's prospectus notes that it anticipates about $40 billion in expenses tied to preexisting stock grants given to employees. These costs have not yet hit the books but will become a factor as the company goes public.

Some observers note that the company's disclosures lack granular details on how specific funds will be allocated. There is a lack of clarity on how much money is going toward training versus serving customers. While some see this as a way to keep the focus on the broad opportunity, others argue it creates vagueness regarding how the company will manage costs moving forward.

The path to a miracle

Despite the financial complexities, Anthropic's leadership maintains a high-level vision for the technology. Chief executive Dario Amodei has predicted that AI could help end cancer and mental illness within a decade. He described the potential for a transformation of cognitive work as the creation of an "age of miracles" in a marketing video.

The company is currently trying to change the public perception of its tools. While many people still view Claude as just a chatbot, the company wants it to be a functional participant in professional environments. They are betting that these tools will become indispensable for white-collar work, creating a massive market for their services.

Investors remain divided on the long-term outlook for the industry. While some believe in the massive opportunity ahead, others are wary of the risks associated with a four-year-old industry. The ultimate success of Anthropic depends on whether it can turn these massive infrastructure investments into a stable, profitable business that delivers on its promises.