AI summary
Summary of article:
Bank of America pointed out that if data center construction is regulated to slow down after the midterm elections, as long as the demand for AI computing power remains strong, the slowdown in supply may actually increase usage, support prices and sustain the growth of chip stocks for a long time.
long term dominancewall Street The idea on AI stocks is that the greater the capital expenditure to develop AI, this will undoubtedly be a good thing for chip makers. Nowdemocratic party in the united statesmidterm elections With China controlling both houses of Congress, AI will face greater oversight and scrutiny. Is the craze almost over? The latest analysis report from Bank of America (BofA) puts forward a different view.
Bank of America analysts pointed out that if data center construction faces regulatory hurdles and slows after the election, AI capital spending will also slow, which may actually prolong the AI boom rather than herald the end of the boom.
This argument of Bank of America is based on one condition. That is, even with the new cooling of data centers, the demand for paid computing services remains stable. Once that becomes true, the analysts wrote: “While demand remains resilient, slower supply growth could boost equipment utilization, supporting prices and reducing the risk of overbuilding.”
His argument is quite simple. Idle servers can barely generate revenue. If the amount of computing power supplied is too large, cloud service providers may need to reduce rates to prevent too much computing power from being idle. This makes it difficult to justify high-cost expansion plans. The current pace of expansion is forced by the political situation to be more cautious, thereby avoiding this trap.
Bank of America said that till now there has been no crackdown on the demand side. According to the latest quarterly financial reports of several cloud giants, the revenue of Amazon’s cloud business AWS increased by 39% compared with the same period last year; Microsoft’s Azure rose 48%, and Google Cloud also jumped a whopping 103%. However, in terms of free cash flow losses, Bank of America estimates that the combined losses of the eight ultra-large-scale cloud service providers and emerging cloud service providers will be US$128 billion this year, increasing to US$335 billion next year, and then decreasing to US$77 billion in 2028.
The market is now pricing in the risks of capex even before cracks in demand data appear. Take Amazon, for example. The company’s latest quarterly financial report showed accelerated growth in its cloud business, but its free cash flow has sharply turned negative $42.4 billion over the past 12 months, partly due to heavy infrastructure spending. As a result, the share price fell after the earnings report was released.
In addition to Bank of America, Monex’s analysis reminds investors that when deploying AI trading, what should be concerned about is not the expense, but the investment return. First there will be capital expenditure, then free cash flow will be squeezed, and then depreciation and power costs will be faced. Therefore, the company’s revenue and profits should be sufficient to bear these burdens. Monex explained that the focus of market judgment is not the size of the investment, but rather how quickly these investments can be converted into revenues and costs recovered.
Monex believes that AI technology lies at the core of the competition between the United States and China. The federal government’s support for AI is unlikely to reverse rapidly due to the midterm elections. However, with the victory of the Democratic Party, the regulations made by states and local governments on electricity, water resources and land use deserve attention.
Monex also raised the possibility that companies could make advance investments ahead of the 2028 presidential election in response to potential changes in federal policy, which would lead to a stronger rather than weaker capital spending cycle in 2027 and 2028.
For stock investors, AI trading is entering a phase where discipline is more important. If the slowdown in manufacturing keeps equipment utilization at high levels, the market will view this as a positive sign for prices and more sustainable chip demand. In comparison, the damage caused by slow demand will far outweigh the impact caused by delays in data center building permits.
Summary FAQ
-
Bank of America believes that if demand remains strong, slowing supply growth will lead to increased equipment utilization, reducing the risk of over-expansion and supporting computing electricity prices, so this does not necessarily mean the end of the AI boom.
-
AWS revenue increased by 39% year-on-year, Azure increased by 48%, and Google Cloud increased by more than 103%, which shows that there is still no clear gap in demand for cloud and AI, at least in the short term to support the corresponding capital expenditures.
-
Monex reminds that the focus is not on how much money is spent, but on how long it takes for the investment to turn into revenues and profits, as capital expenditure will first reduce cash flow and then absorb depreciation and electricity costs.
