If you use generative AI, you know it can seem like magic. Chatbots and multimedia templates can easily conjure up poems or high-resolution videos with the snap of a finger.
But the rapid results and elegant interfaces of AI models mask the enormous amount of physical infrastructure behind them. And as AI continues to grow, the data centers and power plants it’s built on are starting to attract a lot of attention outside the industry.
Earlier this week, I took a train to Orangeburg, New York, a sleepy suburb in the lower Hudson Valley, just 25 miles from Fortunein the downtown Manhattan newsroom. I was there to tour one of the many AI infrastructure projects popping up across the country: Orangeburg is the future home of data center company DataBank’s new facility, called LGA3.
DataBank already operates two data centers in the New York metropolitan area: one in Newark, New Jersey, and one in Chelsea, Manhattan. But LGA3 will be by far its largest site: a 200,000-square-foot facility worth $250 million, consuming up to 45 megawatts of power to power five massive data rooms filled to the gills with chips computers.
The facility won’t open until next year, but tenants have already reserved spaces, including New Jersey-based AI startup CoreWeave, which recently achieved a mind-blowing valuation of $19 billion and has already reserved almost half of the LGA3 capacity.
“The explosive growth of artificial intelligence has required a complete reevaluation of traditional data centers to meet the demand for next-generation computing requirements, and this new data center campus provides some of the most advanced new technologies that will bring us will deliver to our customers,” CoreWeave founder Ben Venturo wrote to me in a note.
My taxi from the station took no fewer than four wrong turns as we drove through farms and office parks to the LGA3 construction site, wedged between an electrical substation and the New Jersey state line. Getting out of the car, my first impression was the size of the building. LGA3 was about the size of a New York City block, a massive single-story lobby with high ceilings. I wouldn’t be surprised if you could put a commercial plane there.
“Addicted to technology”
DataBank CEO Raul Martynek greeted me when I arrived, wearing clear-rimmed glasses and a lavender shirt. Martynek has worked in the Internet infrastructure industry for decades, almost since the advent of the commercial Internet in the 1990s. He has worked at Databank since 2017, overseeing the company’s 69 data centers in the United States and the United States. United Kingdom. Martynek told me he didn’t see an explosion in demand for digital infrastructure like the one that AI has been creating since the Internet bubble of the late 90s.
“Humans are addicted to technology, period. And ultimately, for the data center industry, what we’re doing is enabling humans to deploy more technology,” Martynek told me. “If you deploy more technology, you need more fiber, more cell towers and more data centers. And for this particular phase that we’re in with AI, data centers are the bottleneck.
Nathan Howard/Bloomberg
The huge increase in demand for AI has propelled data centers into the headlines. Martynek explained to me that most of the things we do online these days (from accessing images on our phones to scrolling through social media to prompting ChatGPT) involve more physical hardware than we realize. let’s think. Wi-Fi routers and cell towers constantly send signals through underground fiber optic cables to data centers and remote servers, accessing stored information and operating the Internet.
“The Internet is a network, right? Information is sent via fiber optic cables in the form of photons. And they travel around the world at close to the speed of light,” Martynek said. “Last night I was hanging out with a guy from the network who was like, ‘What are you doing? He said, “We’re plumbers, aren’t we?” »
And these days, being a plumber is a good business. The exponential increase in the amount of data generated for and by the Internet over the past 20 years – and the hope that AI will only speed things up even more – means that the space to store all this information is very demanding.
“This device didn’t exist before 2007,” Martynek told me, pointing to his iPhone. “So think about the amount of content and the number of applications that have been created (by it.) It all ends up in a data center… That’s the physical ecosystem.
Courtesy of SourceCode Communications
AI increases the need for building space
Data centers may not be the sexiest projects, but a surge in demand from AI companies brings big money, heavy media coverage and some of the biggest names in construction. DataBank alone has spent about $4.5 billion on data center projects since 2016. Tishman Speyer, the real estate company building LGA3, is one of the hottest names in the industry: it worked on the World Trade Center and the John Hancock Center in Chicago, and it also owns the Rockefeller Center. A low-slung data bank next to a suburban Little League baseball complex may seem like an odd addition to one’s portfolio, but it’s a safe bet that data centers will prove just as important as skyscrapers. sky.
When I first received the invitation to visit, the location surprised me. New York? Are you home to some of the highest real estate and energy prices in the country? Wouldn’t it be cheaper to build this in the middle of the desert, where land is cheaper and there is access to lower-cost renewable energy?
But Martynek explained that for many clients, it’s simply not practical to be located thousands of miles from one of the most important parts of your business. New York is one of the nation’s largest data center markets, with about 800 megawatts of capacity currently online, much of which is aimed at financial and technology companies that rely on nearby computing capacity to build and trade Round the clock.
“It’s not practical for a data center to be in the middle of nowhere: there’s too much latency,” Martynek said, referring to delays in response time between computers and data centers Offsite. “Too many things can happen along the way.”
“Data centers tend to cluster around metropolitan areas,” he continues. “New York has always been a pretty big market for data centers. It really depends on the population and the companies: If you’re JPMorgan, you don’t want your data center to be in Omaha.
Public policy is also a factor. Kathy Hochul, Governor of New York State unveiled the state’s Empire AI initiative earlier this yearwhich has set aside more than $400 million to finance, among other things, infrastructure such as data center projects.
Donning a hard hat and reflective vest, I walked around the half-built structure with two construction managers. They pointed out the aircraft hangar-sized area where banks of computer chips would eventually be installed, as well as ventilation and water cooling to keep them from overheating.
Once construction is complete, CoreWeave and DataBank’s other customers will begin installing their chips, and DataBank expects the facility to be fully operational by early next year. Once online, CoreWeave will begin renting its computing capacity to tech startups and other AI companies. Martynek told me that DataBank had no trouble finding customers.
“We signed the contract with CoreWeave last year. This building didn’t even exist at the time: it was just dirt. That’s why this product is in high demand,” Martynek said. “There’s a frenzy.”
As the saying goes, strike while the iron is hot: DataBank is already drawing up plans for another site next door, LGA4. Next time you’re around town, keep an eye out for nondescript buildings in your area: the AI data center boom might be closer than you think.
