The data center industry is rapidly expanding to keep up with the growth of the AI flywheel. Although these data centers are the necessary AI infrastructure, they house AI companies' computing, are likely to be expensive to build and even operate, and consume significant amounts of energy. Startups are trying to make data centers more efficient and sustainable, but it's not that simple.
The global data center market is estimated to be worth $301 billion and is projected to more than double to $622.4 billion by 2030, according to P&S Intelligence. Data centers currently consume approximately 4% of the total electricity in the United States. According to the Electric Power Research Institute, it is expected to more than double to 9% by 2030.
Data centers and the large enterprises that rely on them are competing for power. Last month, Microsoft signed a deal with Constellation Energy to restart the Three Mile Island nuclear reactor to meet demand.
Parallel to the rise in data centers, there is also a rise in the number of startups aiming to solve the data center industry's energy crisis and environmental impact. Startups like Incooling and Submer are trying to tap into this space by cooling existing data center technology to reduce heat production. Companies like Phaidra are using software to help manage data center cooling more efficiently.
Some are considering building entirely new models. Verrus is using microgrids to build more “flexible” data centers. Sage Geosystems is building a way to use pressurized hot water to power data centers instead of natural gas.
Sophie Bakalar, a partner at Collab Fund, an investor in Phaidra, told TechCrunch that even before the AI boom, there were entrepreneurs looking to build data center technology, but cloud computing and Bitcoin mining also require data centers. He said it plays a big role. Last year, we noticed a 10x increase in founders looking to build technology for this space.
“We've seen companies building data centers in space, and that's the whole strategy in play,” Bacalar said. “When you have such a clear problem with supply and demand, it’s natural to see a lot of entrepreneurs trying to approach the problem from different angles.”
But while data centers are rapidly expanding and require solutions to increase efficiency, startups should not assume that their technology will be easy to adopt.
Data center challenges
Frances O'Sullivan, managing director at S2G Ventures, told TechCrunch that the fast pace of growth in this space makes it difficult for startups to test their technology and find partners willing to take on the challenge. He said it was possible.
“[Data centers] They are very expensive assets, multi-billion dollar facilities. The reality is they have to work,” O’Sullivan said. “So the real, meaty world of data centers is not a testing ground.”
The customer base for this kind of technology is also probably more concentrated, so penetration will probably be harder, said Christian Blanaes, a partner at climate change-focused VC Transition. Blanaes said his company has spent a lot of time researching and digging deep into the data center technology space, and while they've found great companies building novel technologies, they haven't had enough confidence to invest. He added that it was not possible.
Blanaes is concerned about how companies can scale. He believes some of the startups he's discovered fall into the classic conundrum of climate technology: cool technology, but not necessarily companies that can generate venture-like profits. . He said it's difficult to build a venture-sized company that only sells to a few large companies such as Microsoft and Apple.
“We came to [the] Observation: It's very difficult to build a company large enough to sell to the likes of AWS and Microsoft. They are ruthless in their procurement,” Blanaes said. “They're not in the business of giving you big margins. When you start making too much money, they try to get around it or start doing it internally.”
turn on the power
Although some investors remain skeptical, many startups in this space are gaining traction. With regulations looming both in Europe and in U.S. states where data centers are concentrated, such as Virginia, even if these large customers are not purchasing the solution now, they will have to do so in the future. It means there is a high possibility.
Helena Samodurova, co-founder of Incooling, a Netherlands-based startup aiming to cool data centers, started her company six years ago, predating the current AI hype. Back then, data centers and the energy they used were the issue, but the demands on Incooling's technology have completely changed.
“In the past, people didn't know much about it,” Samodurova said. “Things have changed a lot in the last six years. As we've gone through this journey, we've had to really educate people about what this is. Fast forward six years later, and we're not. They are being searched.”
Mr Samoduoruba said there was also growing interest from both potential customers and investors. He added that the data center industry is much broader than just the Amazons and Googles of the world, and that support for improving data center emissions is not just focused on these few large companies. .
“You have a bus to go to the station, you have a car to take your family out, you have a Ferrari to go to the races, and they all have four wheels, but the mechanisms are different,” Samodurova said. . “We provide a cooling solution or a computing solution to solve any bottleneck.”
O'Sullivan said many data center technologies are a little too early to get excited about at the moment, but other categories should help solve some of the same problems that data center technologies are aiming to solve. He said that there are several companies. . One is to solve the problems associated with delivering real energy to data centers and making sure the power grid can handle that level of power.
For startups focused on data centers, adoption may be too early for some of the first entrants into the category. Unlike Incooling, many of the companies were just founded in the past few years. The data center technology market may still be in its infancy, but the data centers needed to power AI and the industry aren't going away anytime soon.
“I think the main thing to consider is that there is real urgency here,” Bacalar said. “That growth is actually outpacing the current infrastructure that we have. We need newer, better, and faster ways to deliver on the promise we've heard about AI.”