US tech giants maintain high levels of AI investment

Amazon and other four companies will increase their equipment investment by 60% in 2024 to about $245 billion, and will maintain a high growth of 30% to 70% in 2025. Concerns about the huge investments of US IT giants have intensified due to the debut of DeepSeek in January.

US IT giants will continue to make huge investments in artificial intelligence (AI) infrastructure. Equipment investment of four major companies including Amazon in the United States will reach about $245 billion in 2024, and is expected to remain at a high level in 2025. Even in the context of the rise of Chinese start-up DeepSeek, which has made the return on investment worrying, companies still maintain an offensive posture.

The financial reports of the four major US IT companies involved in AI data centers (Google’s parent company Alphabet, Microsoft, Amazon, and Meta) from October to December 2024 were all released on February 6. The equipment investment of the four companies from January to December 2024 increased by 60% year-on-year to about $245 billion.

Active investment will continue in 2025. Amazon’s equipment investment in 2025 is expected to increase by about 30% from the previous year to about $100 billion. Meta is expected to grow by about 70% to $65 billion, while Alphabet is expected to grow by about 40% to $75 billion. Microsoft will invest $80 billion in data centers in fiscal year 2025 (ending June 2025).

Most of the equipment investment is in chips used for AI computing and processing in data centers. High-performance chips mainly produced by Nvidia in the United States are embedded in servers to allow them to learn large amounts of data to improve the accuracy of AI.

Dell’Oro Group, a US research company, predicts that the investment in AI infrastructure, that is, data centers, by the four major companies will reach about $540 billion by 2029, nearly three times that of 2024.

Although they are all in the IT industry, the profit sources of companies such as Alphabet and Meta, which are centered on advertising, Microsoft, which is involved in software, and Amazon, which is involved in cloud computing and e-commerce, are completely different. But when it comes to developing AI, the attitudes of the companies are consistent.

On the other hand, the profit model of AI business is different. Amazon, Microsoft and Google hope to make money through AI by providing cloud services of IT infrastructure. The basic technology of AI will be provided through the cloud platform, and customers can use AI to develop related services. This is a strategy to promote the use of cloud platforms through AI services.

For cloud computing companies, the effect of AI investment is also reflected in profits. From the perspective of Microsoft’s main cloud computing platform “Azure”, AI-related operating income from October to December reached 2.6 times that of the same period last year.

Meta is exploring a different way of making money from cloud platform companies. Meta is deploying AI on existing products and services such as social networks (SNS) and glasses-type terminals, and is improving ease of use.

Concerns about the huge investment of US IT giants have intensified due to the debut of DeepSeek in January. DeepSeek announced that it has developed low-cost AI that can compete with the most advanced technology. As concerns spread that a large number of chips used for cutting-edge AI development would no longer be needed, Nvidia’s total market value evaporated by more than 90 trillion yen.

Despite this, US IT companies still maintain an active investment posture. The idea is that with the advent of low-cost AI, the entire market will expand.

Amazon CEO Andy Jassy said at the earnings conference on the 6th, “Typically, if the unit price of technology drops, overall spending will increase. I think the same thing will happen in the field of AI.” Microsoft CEO Satya Nadella said about the development of low-cost AI, “This is good news for our company.”

Not only when developing AI technology, but also when using AI services, IT infrastructure such as cloud computing is required. Of Nvidia’s AI chip revenue, about 60% is for development purposes and about 40% is for use services.

Even if the method of developing AI at low cost spreads, as long as the use of AI increases, the revenue obtained through the cloud platform will also increase. As long as this pattern is not broken, investment in AI is likely to continue.

However, there is no guarantee that the huge investment will be recovered in the long run. All three cloud computing companies failed to meet investors’ expectations in this business, and their stock prices fell after the earnings were announced. If the assumption that profit opportunities will increase as the unit price of AI drops fails to materialize, the stock market may raise doubts about U.S. companies that have been investing heavily in the technology.