Written by Joy Wiltermuth
Forget glittering skylines, the new bright spot in commercial real estate is workforce housing for computers that can feed data into artificial intelligence.
The newest hotspot for commercial real estate isn't in Manhattan or Miami.
Instead of fancy hotels and glitzy office towers, the new real estate darlings are power-hungry data centers, often located in places like Northern Virginia. Columbus, Ohio. and Salt Lake City.
With a traditional boxy style, these buildings are all about function. A place where racks and racks of computers are stacked high, kept cool, and guarded against an endless stream of images, videos, chats, texts, internet searches, and digital detritus. our life.
Increasingly, their goal is not simply to contain, host, and classify the data that is sent to them in large quantities by computers and smart devices, but to provide a place where machines can learn from the data.
“Everything is now AI,” said Sean Farney, a data center industry veteran and vice president of data center strategy at real estate firm Jones Lang LaSalle. “The most immediate impact will be a huge new natural demand to host the technology that powers AI.”
Farney, who helped Microsoft open its then-largest 120-megawatt data center in Chicago 15 years ago, said the release of ChatGPT in late 2022 marked a turning point for the industry.
This has allowed companies like Microsoft (MSFT), Google's parent company Alphabet Inc. (GOOG), and Meta Platforms Inc. (META) to deliver their artificial intelligence products more quickly. It also fueled a gold rush in a niche area of commercial real estate that would otherwise have suffered, particularly when it came to office buildings.
AI that needs power
ChatGPT, which reached an estimated 100 million monthly users in its first two months, can generate text, images, and more. Wall Street's enthusiasm for all things AI has provided rocket fuel to the stock price of chipmaker Nvidia Inc. (NVDA) and other “Magnificent Seven” technology stocks.
Listen: Nvidia's journey from video game darling to AI powerhouse
Server maker Super Microcomputer (SMCI) has been one of the stocks to benefit recently. And AI is also focused on boom-ready real estate.
“It's no longer a niche,” said Patrick Wilson, a portfolio manager who has specialized in data centers for more than a decade at Center Square Investment Management, a global real estate investment firm.
Mr Wilson said that previously a brutal approach to new supply would have “hit landlords' feet”, but pricing power has clearly shifted in favor of data center owners in the wake of the pandemic.
In recent years, companies have competed to lease data center space in primary markets such as Northern Virginia, and secondary markets such as Amazon.com's $7.8 billion development in central Ohio. Demand is also spreading to emerging markets. .
CenterSquare's Wilson said that since 2020, when remote work took hold, there has been an incredible amount of capital flowing into opportunities to enable data centers, communications infrastructure, fiber, cell towers, and related technologies. . “There is exponentially more money flowing into this area than any other area of commercial real estate,” he said.
Much of the funding comes from well-funded hyperscalers, including many of Wall Street's Magnificent Seven stocks, and from Oracle Corporation (ORCL), a major high-tech company that builds its own data centers on a global scale. The flow is coming from companies like . The compound annual growth rate is 20% to 30%, according to JLL's Fernie. They are estimated to have spent as much as $1 billion on the new facility.
All told, JLL (JLL) tracks more than 5.3 gigawatts of data center capacity under construction in North America, or “enough energy to power every household in the San Francisco metropolitan area for a year.”
land, water and power
The race to power AI is forcing developers to seek out ever more valuable land with the right natural resources and permits.
For existing facilities, the focus is on upgrades, redesigns, and alternative cooling technologies. However, JLL says new AI-focused data centers often have smaller physical footprints and can be built in more remote locations, provided there is sufficient power.
Finding available power and land in municipalities open to zoning development projects is what Jackson Garton, co-chief investment officer of Menlo Park-based Makena Capital Management, and his team believe is the final This is an area that we are currently considering with a view to selling it to a larger company. players, companies or infrastructure funds.
Makena's roots are in Silicon Valley, and it was founded nearly 19 years ago as a spinout from an endowment at Stanford University. It currently manages approximately $22 billion in assets.
“The demand for data consumption and external processing of data will continue,” Garton told MarketWatch. He also predicts that data centers will likely eventually rival industrial, hospitality and other commercial real estate cores.
However, growth will not be easy as power and land constraints are already major obstacles for the industry. In the latest sign of backlash, Google's permit for a $200 data center planned for Santiago, Chile, but not yet built, was partially revoked by an environmental court this week. Google is asking Google to amend its application to take into account the effects of climate change. Changes especially taking into account the capital's water supply.
Google said in a statement to MarketWatch that it had already made the design change in February 2022 “in line with its commitment to climate-smart data center cooling” and “uses air-based cooling.” “We will continue to cooperate with requests from local governments.”
The company also said sustainability is a core focus, including in its data centers. In its January earnings call, Google ranked data centers as the second-largest component of its $11 billion in total capital spending for the quarter.
Can't build supply fast enough
“Competition for data center space is fiercer than ever, with properties in the early stages of development consistently being pre-leased,” Ermengardo Jabir, senior economist at Moody's Analytics, wrote in a February report on AI and commercial real estate. “It has been done,” he said.
Jabir said office-to-datacenter We believe that there is a possibility that this transformation will progress further.
Still, Jabir said the needs of AI will put additional strain on an “overstretched” infrastructure, noting that AI uses a lot of power to operate and water to keep it cool.
JLL's Farney said today's hyperscale data centers consume as much power as cities. But new servers running AI applications like GPT require five to 10 times more power, he said.
That means using more efficient designs, layouts and materials, and reusing other buildings to feed the growing AI beast, especially as the sector grows in a more environmentally friendly direction. Farney said that likely meant .
“AI factory”
Big tech companies are the obvious owners of data centers in the United States, but they're not the only ones. Their needs are so great that they rent space from others.
Digital Realty Trust Inc. (DLR), Equinix Inc. (EQIX) and other real estate investment trusts are among the established players that have been in the space extensively over the years, in part because customers have been reluctant to give up. It has become. We declined a location or long-term lease due to concerns that it could disrupt our day-to-day business operations. These existing players will need to find ways to compete in the AI gold rush while dealing with traditional data center assets that are at risk of obsolescence.
In recent years, major infrastructure funds such as Brookfield Infrastructure Partners and Blackstone's funds have joined the trend, making a series of splashy investments in data centers as global demand soars.
“We believe it will take more than 10 years to build AI, and the world's current data center capacity is 50 gigawatts, and we plan to double that capacity to 100 gigawatts over the next six to 10 years,” said DigitalBridge. Group's Mark Ganzi said. said the CEO at an earnings conference in February.
The company, which invests in cell towers, data centers and other digital infrastructure, had approximately $80 billion in assets under management at the end of 2023 and expects further growth in this area.
“As we see data centers becoming AI factories, with data as input and intelligence and insights as output,” Ganzi said.
-Joy Wiltermuth
This content was generated by MarketWatch, a Dow Jones Company. MarketWatch is published independently of the Dow Jones Newswires and the Wall Street Journal.
(Ended) Dow Jones News
03-02-24 0544ET
Copyright (c) 2024 Dow Jones & Company, Inc.