Как энергетические затраты ставят под сомнение перспективы ИИ-компаний Translation: How Energy Costs Put AI Companies Futures in Question

Current revenues for AI companies may not justify the substantial costs associated with computing. This statement was made by HSBC CEO Georges Elhedery at the Global Investment Summit for financial leaders in Hong Kong, as reported by CNBC.

In July, Morgan Stanley analysts noted that the capacity of global data centers is projected to increase sixfold over the next five years, with the valuation of data centers and their equipment expected to reach $3 trillion by the end of 2028.

A McKinsey report published in April estimated an even larger figure—by 2030, meeting the demand for AI infrastructure will require capital expenditures of $5.2 trillion. Spending on data centers supporting traditional IT applications is expected to stand at $1.5 trillion.

According to Elhedery, consumers are reluctant to pay for this, and companies will exercise caution since performance advantages may not materialize for a year or two.

«It resembles five-year trends, and thus growth means we will start to see tangible benefits in terms of revenue and readiness to pay, likely later than investors expect,» he remarked.

William Ford, the Chairman and CEO of General Atlantic, concurred with this assessment:

«In the long run, you’ll create a whole range of new sectors and applications, delivering benefits in terms of enhanced productivity, but it will take 10-20 years.»

Major tech companies like Alphabet, Meta, Microsoft, and Amazon have raised their capital expenditure forecasts to $380 billion by 2025. OpenAI has entered into several infrastructure agreements worth about $1 trillion.

Ford highlighted that the enormous expenses in the AI sector demonstrate an understanding of the technology’s long-term impact. However, it’s necessary to «pay upfront for the opportunity that will arise in the future.» At early stages, there may be «misallocation of capital, overvaluation, and irrational enthusiasm.»

«You are essentially betting that this will be a wide-scale technology, more akin to railroads or electricity, which have had profound impacts over time and altered the economy. But for the first few years, it is quite challenging to predict how exactly that will unfold,» concluded the CEO of General Atlantic.

Recently, investors have been actively discussing whether the markets are overestimating artificial intelligence.

Last week, investor Ray Dalio stated that his personal «bubble indicator» is at a relatively high level. At the same time, Federal Reserve Chairman Jerome Powell characterized the AI boom as «distinct» from the dot-com situation.

Magnus Grimeland, founder of the Singapore venture firm Antler, believes that the industry is «definitely» not in a bubble. He noted that the pace of integrating neural networks into business is faster compared to other technological changes, such as the shift from physical servers to cloud computing.

Moreover, artificial intelligence is a «priority task» for thought leaders, whether they are heads of medical institutions in India or executives from Fortune 500 companies in the United States.

«What distinguishes this situation from a bubble and makes it entirely different from the dot-coms is that much of the growth is supported by real revenues,» Grimeland said.

Another factor setting AI’s popularity apart from the dot-com boom is the speed of consumer adoption.

«Think about how quickly our online behavior has changed, right? A year ago, 100% of my search queries were on Google. Now, it’s probably down to around 20%,» Grimeland mentioned.

AI projects are increasingly being integrated with familiar online systems. In October, OpenAI introduced its Atlas browser, featuring an integrated chatbot and AI assistants.

OpenAI’s annual revenue exceeds $13 billion, according to company CEO Sam Altman in a podcast. While this is a significant amount, it pales in comparison to the $1 trillion the startup intends to spend on computing infrastructure over the next decade.

Host Brad Gerstner posed a question to Altman about this, to which he responded:

«First of all, we are acquiring much more. Secondly, Brad, if you want to sell your shares, I can find a buyer for you.»

He added that there are critics who «excitedly discuss computational systems or something else and would be willing to buy shares.»

Altman acknowledged that there are scenarios that could lead to issues, such as a lack of access to sufficient computational resources. However, «revenues are growing rapidly,» he added.

«We are betting on continued growth, and this applies not only to ChatGPT. We hope to be one of the key AI services, our consumer device manufacturing will be significant, and artificial intelligence capable of automating science will generate tremendous value,» remarked the entrepreneur.

Microsoft CEO Satya Nadella noted that OpenAI has «outperformed» all business plans submitted to his company as an investor.

Major corporations continue to invest billions of dollars in AI initiatives, despite discussions about a potential bubble in the sector.

In April, OpenAI secured $40 billion in funding at a valuation of $300 billion. In October, the company allowed current and former employees to sell shares worth $6.6 billion. In this deal, the startup’s valuation reached $500 billion—a record among private firms.

On November 3, cloud computing startup Lambda announced a multibillion-dollar agreement with Microsoft to develop AI infrastructure based on tens of thousands of Nvidia chips.

«We are in the midst of perhaps the largest technological buildout we’ve ever seen. The industry is currently flourishing, and many people are using ChatGPT, Claude, and other available AI services,» commented Lambda CEO Steven Balaban.

On October 3, Microsoft announced an investment of $15.2 billion in the UAE over four years. This includes supplies of advanced Nvidia graphic chips.

As part of the agreement, the United States granted the corporation a license to export chips.

The company has begun spending investment funds in the region since 2023. The new agreement entails $7.9 billion in investments from early 2026 through the end of 2029, including $5.5 billion for capital expenditures and AI infrastructure expansion.

Microsoft also struck a $9.7 billion deal with Australian company IREN to provide cloud computing capabilities for artificial intelligence. The agreement will grant the corporation access to infrastructure built on Nvidia GB300 graphic processors.

Grimeland emphasized that «huge» amounts of money are flowing into AI-related companies due to «mispricing,» but the opportunities in the field are far greater.

Energy is one of the main drivers of the massive expenses associated with artificial intelligence. The operation of hundreds of thousands of graphics cards requires a constant flow of electricity. This puts pressure on the grid and leads to rising utility prices, which is a concern for end consumers.

How much energy will be sufficient for AI? No one knows, not even Altman or Nadella.

«In this particular case, cycles of supply and demand are genuinely unpredictable. The biggest challenge we are currently facing is not an excess of computing power, but the ability to build [data centers] quickly enough near power sources,» stated Microsoft CEO in the podcast.

Otherwise, the company may end up with too many chips in storage without a way to connect them, he added.

The surge in electricity demand has outpaced the plans of utility companies to create new generating capacities. This has led data center developers to secure energy bypassing the grid through special contracts.

«If a very cheap form of energy becomes widely available soon, many will be extremely disappointed with existing contracts they signed,» Altman pointed out.

It should be noted that in July 2024, Bernstein Research suggested the possibility of an energy shortage in the United States if the demand growth from AI data centers continues at its current pace.