How Big Tech plans to meet AI's energy demands through nuclear and quantum computing

• The global increase in data centers is not slowing down, leading Big Tech to contemplate how to power the AI revolution.
• Some of the options on the table include a shift to nuclear energy, liquid cooling for data centers, and quantum computing.
• Raj Hazra, CEO of Quantinuum, advised CNBC that every great technology has a winter season, but it's essential not to notice it until it arrives.

The global increase in data centers is not slowing down, leading Big Tech to contemplate how to power the AI revolution.

Some of the options on the table include a shift to nuclear energy, liquid cooling for data centers, and quantum computing.

The "move fast and break things" narrative should be abandoned by tech giants as critics argue that the pace of efficiency gains in electricity use is slowing, and the cost of the generative AI boom should be recognized across the whole supply chain.

The true environmental cost is currently concealed, as it is being subsidized by the need of tech companies to launch a product and secure a buy-in, according to Somya Joshi, head of division: global agendas, climate and systems at the Stockholm Environment Institute (SEI), who spoke to CNBC via video call.

The International Energy Agency predicts that the investment in data centers will increase at a faster rate in the near future, mainly due to the increasing digitalization and adoption of generative AI.

The prospect of increased electricity demand has sparked concerns about the potential surge in energy consumption, as well as the environmental impact of AI.

Cloud computing and AI applications rely heavily on data centers, which consume an increasing amount of energy.

ABB's data center business has experienced significant growth in recent years, with the segment projected to increase by more than 24% in 2024, according to Giampiero Frisio, president of electrification at the Swiss multinational.

ABB has been positioned to provide data center components to mid-sized and large industry players during the AI demand surge, as stated by Frisio.

Frisio advised CNBC via video call that the best course of action now is to increase energy efficiency, as the technology is available, such as the medium voltage HiPerGuard UPS. This can be done, and it can be done tomorrow morning.

ABB's HiPerGuard UPS is an industry-first medium voltage uninterruptible power supply that can provide continuous power to large facilities.

Frisio stated that the second option is liquid cooling, which is more energy-efficient. He explained that this is because a single rack, which resembles a wardrobe with servers inside, has a power density of four to six times more than before.

He stated that in five to 10 years, the nuclear modular system would be the topic of discussion.

Big Tech is going nuclear

In recent months, Microsoft, Google, and Amazon have secured nuclear energy deals worth billions of dollars to power their massive generative AI models.

As the demand for generative AI increases, there is a growing need for more efficient cooling solutions in data centers, specifically liquid cooling, which involves using water to cool servers and other electronic equipment.

Schneider Electric recently acquired a controlling stake in Motivair Corp, a U.S.-based company specializing in liquid cooling for high-performance computing, in an $850 million deal.

Last month, Peter Herweck, CEO of Schneider Electric, stated on CNBC that the all-cash deal aimed at enhancing its data center offerings was "expensive but not excessively so" and aligns well with the company's strategy.

Some tech companies have proposed using AI to reduce carbon emissions in data centers, in addition to nuclear energy and liquid cooling technology.

Last month, former Google CEO Eric Schmidt stated that since "we won't achieve our climate targets anyway," investing in AI could be crucial in addressing some of our most significant environmental issues.

SEI's Joshi flatly rejected this point of view.

Joshi stated that the arguments being presented are not original and align with the typical "silver bullet" and "tech will save us" narrative.

"By exceeding the planetary boundaries we operate within and continuing with the same extractive narratives, we are somehow going to solve the problem we're currently facing," she stated.

Quantum computing

Raj Hazra, CEO of Quantinuum, stated in a video call with CNBC that he believes there is a winter season for every great technology discovery in the summer, but it is essential not to pay attention to it until winter arrives.

"My way of describing the current state of generative AI involves the infrastructure required to support it, including the construction of massive data centers."

The cost of running generative AI technology is becoming increasingly expensive due to the growing optimism about its potential.

"While AI has many benefits, two challenges remain: sustainability from resources and responsibility. Quantum technology is crucial to addressing these challenges."

Quantum computing utilizes quantum mechanics to tackle highly intricate problems.

Quantinuum, a company that has recently raised $300 million in its latest equity funding round, has attracted the interest of various businesses and strategic investors, including Honeywell, which is its majority shareholder. The fundraise has put Quantinuum's valuation at $5 billion.

"Sustainable solutions are now preferred over saying 'I have a solution,' as pointed out by Hazra," said Hazra.

One of quantum's greatest contributions to society is ensuring that AI is sustainable and responsible.

In the next three to five years, people will ask about their compute infrastructure for running their business, and the answer will involve high-performance computing, AI, and quantum, he stated.