Generative artificial intelligence explodes the demand for data centers, resulting in a multitude of operational and energy challenges. Proptech companies are trying to offer solutions – including the use of Genai itself to remotely control the operations of the data center.
The demand is so large – and the energy consumption of the data centers so voracious – that the installations formerly measured in square feet or by the number of racks they could adapt are now measured by their electricity consumption, said Peter Hannaford, CEO of EdgenebulaA London supplier of sustainable cloud and AI -focused technology for data centers.
“They are measured by the megawatts and more and more in Gigawatts,” said Hannaford in an email. “Thus, the major data centers are under construction where the power is located. Proptech must be ahead of the game or be ready to adapt to energy production on site – and possibly storage when technology is available. »»
Citigroup estimated last May The size of the data centers market was 33 gigawatts in 2023 and predicted annual growth made up of 17% to 100 gigawatts in 2030.
Despite the fears that China In depth Would have a negative impact on the growth of the American data center, the reverse occurs. Technological expenditure of the data center A Zoomé 34% in 2024 at $ 282 billion, according to Synergy Research Group. It will have exceeded half a bill of dollars during the first month of 2025.
Among the great players of cloud composed, Amazon AWS announced on January 7 that it was Invest $ 11 billion In her georgia facilities after saying last year that she would spend $ 100 million During the next decade on various places. Furthermore, Google CloudGoogle’s cloud computing activity last year said it had a Annual execution rate of $ 37 billionwhile Meta announced that he is investing $ 65 billion in 2025 alone.
Narpetful numbers.
However, the marketing period is a major problem in building data centers, often requiring two years from start to ignition, said Hannaford. Like other companies, Edgenebula Built for himself smaller 500 kilowatts data centers to use the many locations where electricity is available in the middle of an abundant property supply rich in customers in and around metro locations. He advises other Proptech companies to readjust their portfolios accordingly.
While the data centers increase in number and complexity, fewer installation staff will be necessary on site, and even remotely, said Hannaford. In other words, the day is fast approaching when the data centers, even far from the population centers, could be exploited almost entirely from a distance.
“There is a popular false idea that the construction of these new massive data centers will create the creation of new jobs,” he said. “This is only true during the construction phase. Once built, there are far fewer people employed in a data center than in a commercial building. Data centers are designed for machines, not people. »»
The growing need to remotely control data centers focuses mainly on their energy consumption, but also on security requirements, he continued.
“If the data center occupies part of a larger building, interfaces with buildings management systems will also be necessary,” said Hannaford. “If the data center is part of a larger national or international network, the connectivity of fibers between data centers will be necessary.
“With pressure on operators to reduce their carbon footprint from governments and certainly customers, remote monitoring of efficiency, effective use of energy and effective cooling systems are essential. In addition to efficiency, Proptech’s main role will be to offer resilience – also almost 100% possible. Thus, remote control systems are necessary to switch to emergency systems, including power, cooling and connectivity in the event of failure. »»
What Proptech can add to support the mega-growth of the data center is “the question of the billion dollars”, Ajey Kaushal, director JLL Spark Global Venturessaid in an email. “Although Proptech plays a substantial role here, he does it in concert with many other verticals. He offers solutions that help solve this problem inside and outside the box. »»
According to Kaushal, certain examples of innovations inside the boxes Flow And Coolgradent; “Loading charges” to ensure that critical systems receive power during the peak request via companies such as JCI OpenBlue And Brainbox Ai; and actively monitor the quality of energy through Phaidra,, Ekkosense and similar companies.
As for the outside of the trails Rhythmswhich helps to identify the plots of land conducive to the development of the data center. Other solutions include processes and materials to be built more quickly and more efficiently via companies such as Lobby And Bending; Management of the electricity production and site storage infrastructure through companies as Edgecentres; And help in capital planning for an increase in electricity production or expansion of the data center, for example by rhythms and Pearl Street. (JLL Spark is not invested in these companies.)
Although it is not limited to data centers, the Proptech startup Phaidra Provides such locations an AI platform which includes remote controls, said Jim Gao, co-founder and CEO of the company based in Seattle. Phaidra provides an AI control service in a closed loop that helps operators maintain plant stability, energy efficiency and sustainability in critical mission facilities of all kinds, he explained.
“We use a type of AI known as the learning by strengthening, which is one of the four large branches of automatic learning,” said Gao. “It is also the only type of automatic learning that learns by doing, like humans. He learns the same way by having software defined agents who take measures in their environment. In our case, they will take measures such as modifying variable frequency driving pump speeds or lighting and deactivating coolers. And they learn from their actions to better manage the installation, which means more reliable, safer and more energy efficient. »»
Mechanical engineer, Gao is delivered by his knowledge of the AI and the data center for his previous work at Google, where he was a successful engineer and produced Energy efficiency gains of 30%. In 2019, he left Google with Google Deepmind The main engineer Veda Panteershelvam, now director of technology at Phaidra, and Katie Hoffman, formerly of Trane TechnologiesAnd now the chief of the Phaidra exploitation.
The company has closed a series of a series of around 60 million dollars in 2024, and now has nearly 100 employees, with customers in the United States as well as in Australia, Canada, Europe, Singapore and In the United Kingdom, according to society.
“The main value is energy savings and improvements in reliability via AI control,” said Gao, who refused to appoint customers and the number of companies with whom Phaedra worked. “We work with a large part of the largest corporate corporations of data centers in the world, and increasingly with large technological companies, mainly in the United States, although we are active with companies in the UE, as well as with Southeast Asia, with a very the accent on Singapore in this region. »»
The exponential increase in the technical complexity of the data center may not be as obvious as its energy needs, Gao said.
“The data centers are becoming increasingly large. There are more and more machines that go there, “he said. “There are more and more complex workloads that come there where, instead of very stable (central processing units), you see high density processors with liquid cooling at the maximum workload. So you have this order of magnitude in complex installations.
“And this is where AI really shines, because AI is good for analyzing and managing massive complexity, as long as it is supported by the data, of course.”
Philip Russo is reached at prusso@commercialobserver.com.