The world’s largest technology companies are aggressively increasing spending on AI data centers to meet growing demand. However, Wall Street’s expectations for a quick return on billions of dollars in investments have not yet been met. Amazon, which will report on Thursday its results for the third quarter of this year, is likely to confirm forecasts regarding the high costs of AI development.
Microsoft and Meta✴, along with other tech giants, continue to increase capital investments in AI and data centers. On Wednesday, both companies reported significant increases in spending directly related to the development of AI infrastructure. Alphabet also confirmed on Tuesday that its costs will remain high in the near term to maintain capacity for AI solutions.
Large capital expenditures are threatening the tech giants’ high margins, and the pressure on profitability is likely to add to investor concerns. Although Meta✴ and Microsoft beat analysts’ expectations for third-quarter revenue and earnings, shares of both companies fell 4% in Thursday’s pre-market trading. Amazon shares also fell 1.4%, reflecting concerns from shareholders who expect the companies not only to excel in AI but also to have relatively short payback periods on investments.
According to GlobalData analyst Beatriz Valle, maintaining and developing AI technologies is expensive, and access to the necessary capacity is becoming increasingly expensive. Major technology companies have entered the race to lead the way in building AI computing power. Microsoft’s AI infrastructure investment in the quarter alone now exceeds its annual spending through 2020, according to Visible Alpha data. Meta✴ also significantly increased its expenses: the company’s quarterly expenses reached a level comparable to its annual expenses before 2017.
Microsoft has warned that growth in its key Azure cloud business could slow due to capacity constraints in its AI data centers. Head of Technology Research D.A. Davidson Gil Luria noted that Microsoft’s annual overinvestment, similar to what the company is making this year, would reduce margins by 1% over the next six years. Meta✴, in turn, also warned that its spending on AI infrastructure will increase significantly next year.
The problem of lack of capacity for AI is also relevant for chip manufacturers. For example, Nvidia, a leader in high-performance graphics processing units (GPUs), is finding it difficult to meet growing demand for its products, limiting the ability of cloud companies to expand their infrastructure. AMD also confirmed this week that demand for its AI chips is growing faster than production capacity and warned that shortages of AI chips will continue into next year, making it difficult to fill all orders.
Despite the current challenges, Meta✴ and Microsoft emphasize that AI technologies are only at an early stage in their cycle and their potential for the future is enormous. These strategic investments in AI infrastructure are reminiscent of the early days of cloud technology, when companies also expected a long process of customer adoption of new solutions. Meta✴ CEO Mark Zuckerberg noted on an investor call: “Building AI infrastructure may not be what investors want to hear in the short term, but I think the opportunity here is really big.”