Flexential CEO Chris Downie outlined a brand new playbook for information heart operators, amid unprecedented demand, funding and chokepoints
RESTON, Virginia—The wake-up name for Flexential CEO Chris Downie began innocently sufficient: He despatched an image of himself assembly former Ukrainian heavyweight boxing champion Wladimir Klitschko to a good friend, by which each he and Klitschko had been smiling and good-naturedly flexing for the digital camera. Inside quarter-hour, the good friend despatched again a short video, seemingly of that very same second: The boys grinned, flexed — then turned to one another, squared off and started sparring.
The sparring by no means occurred. The video was AI-generated from Downie’s image. However for Downie, the surreal expertise of realizing that AI might flip his picture into something, create moments that appeared actual however had been utterly fabricated — and he had no management over it — was emblematic of simply how far off the map that the info heart trade, the place he has spent twenty years, was headed.
“The machines are re-writing actuality in actual time,” he instructed the viewers at this week’s Knowledge Heart Frontier Tendencies Summit. “And in the event that they might rewrite my face and the scene of which I’m in, in lower than quarter-hour, what’s it going to do to the trade?
“Clearly, AI is already rewriting actuality. It’s transferring markets and momentum that we haven’t seen earlier than,” Downie continued. Whereas some are framing the expansion as simply one other cycle of boom-and-bust — and Downie took care to run via quite a lot of the up-and-down cycles the info heart trade has already weathered — he thinks this time is essentially totally different. “From my perspective, this isn’t short-term. And that is the second that’s going to redefine our position as operators, as infrastructure enablers within the trade,” he mentioned.
Knowledge heart operators have roughly adopted the identical playbook in previous cycles, he mentioned, tweaking and adapting based mostly on classes discovered about issues like energy volatility and provide chain points, with expertise informing how new challenges can be approached. “However this new cycle invalidates previous expertise,” Downie mentioned, including: “As we glance ahead, we have to respect that the whole lot’s totally different. And doing issues enterprise as common, the identical outdated means, isn’t going to use.”
So what are the brand new guidelines for information heart operators? Downie outlined the foremost trade touchstones which might be being utterly reshaped.
–Energy wants. That is maybe the most blatant crunch throughout the info heart trade. Downie cited numbers from Reuters that indicated the common load from AI and information facilities this yr is projected to hit 488 gigawatts — the best ever. Provide shouldn’t be conserving tempo, with annual demand outstripping utilities’ potential to offer. The hole between what the grid can present and what information facilities want might exceed 20 gigawatts by 2030, he warned, which creates cascading challenges for DC operators.
These cascading results embrace utilities shifting prices to information heart operators in ways in which they haven’t beforehand, and demanding up-front funding. “That was not within the outdated playbook,” he mentioned. “So when you consider constructing substations or transmission traces, you now have to consider doubtlessly a whole bunch of tens of millions of {dollars}, upfront of deployment.” That was once a price that was amortized over time as a part of the enterprise mannequin.
–Land getting locked up. In the meantime, hyperscalers, and what Downie referred to as “hyperspeculators”, are snapping up properties that would doubtlessly function information heart websites years out, or perhaps a decade out, from any precise constructing plans. “That’s going to trigger challenges for all of us. … We’re going to must advance capital, to get land nicely upfront of after we in the end wanted to prior to now,” he mentioned.
–Construct timelines preserve stretching additional and additional out. Within the outdated playbook, maybe the timeline mannequin was 5 years till a DC began producing return. Now, it is perhaps longer than 5 years simply to get the infrastructure constructed; in lots of instances, the wait time for energy grid connections are exceeding seven years. If transmission tasks are concerned, the wait may be even longer. That degree of delay, Downie mentioned, could also be “going past the horizons which might be doubtlessly set by the buyers which might be fueling the capital to drive our infrastructure.”
–Provide chains are stretched. Connection to energy isn’t the one bottleneck. Crucial gear like distribution transformers can face lead occasions of as much as 5 years, in line with Downie. “That delay will influence even the best-prepared operators,” he mentioned. “And we’re not speaking about simply transformers. You’ve bought swap gear, turbines, cooling techniques … bodily plant has develop into a strategic problem. So, the brand new playbook will power us to suppose extra dynamically about planning, procurement, and constructing than maybe we’ve ever skilled prior to now.”
–Constraints on elasticity. Up to now, information heart operators might assume that there was some predictable elasticity in provide, the availability chain, compute, expertise and the essential uncooked supplies wanted for constructing and working information facilities. Issues, and other people, had been accessible. They might be procured. That elasticity “isn’t accessible to us anymore. It actually must be engineered,” Downie mentioned.
–Infrastructure going from passive to pivotal. Digital infrastructure, Downie mentioned, “has moved from the boiler room to the entrance traces.” That’s a brand new place for the trade. “For 17 of my 20 years within the enterprise, information facilities had been within the background. Quiet, dependable, secure, however … in some ways backstage.”
Now, it’s a strategic differentiator — “how innovation will get deployed, scaled, and monetized,” as Downie put it. Meaning the info heart operators are transferring from being primarily “order takers” to “innovation enablers.”
–Capital isn’t equally distributed, and prices are excessive. The frequent assumption is that the billions of {dollars} pouring into the info heart market will tackle the calls for of hyperscalers. However the place does that go away the smaller enterprise clients who’ve crammed information facilities for the previous twenty years? Their wants aren’t going away, they usually need to undertake AI instruments as nicely — however now there’s extremely intense competitors for information heart assets. Downie mentioned that he has needed to resolve whether or not to allocate information heart capability between two loyal Flexential clients, as a result of there merely wasn’t sufficient for each.
“The enterprise wants to start, if not proceed already, to be excited about the way it thinks about elasticity in its personal new playbook,” Downie suggested. Meaning excited about procurement timelines and being lifelike about prices. Within the new playbook, Downie sid, “All the things is pricey. Land, energy, labor, persistence. … All of those parts are new currencies within the new playbook. And … none of us on this room, although, are within the enterprise of absorbing prices to allow our clients’ development.”
The brand new playbook is outlined by shortage
Amid the expansion, he mentioned, a wide range of assets are being examined by shortage. “We’re not simply speaking about megawatts,” Downie mirrored. “We’re speaking about land, energy, group relations, touchstone expertise — most of the uncooked supplies that assist the expansion of the enterprise.”
Flexential’s annual state of AI infrastructure report, based mostly on a survey of 350 enterprises about their plans for AI, confirmed that survey individuals are feeling extra assured about AI than they had been a yr in the past. “Issues are undoubtedly accelerating. … There’s excessive confidence that AI roadmaps should be developed. However the fascinating factor is that virtually half of these enterprise IT leaders say that infrastructure is the best barrier to scaling AI.”
The Flexential CEO described the brand new setting as one outlined by shortage, the place leaders might be judged by their potential to make laborious trade-offs. He acknowledged having to prioritize capability for one buyer over one other in latest months, a choice he by no means beforehand contemplated. “Shortage goes to separate the leaders from the loud,” he mentioned.
Regardless of the challenges, Downie remained optimistic in regards to the trade’s future. He argued that infrastructure should not solely preserve tempo with AI but additionally be acknowledged as “an engine of financial development, nationwide competitiveness, and innovation at a worldwide scale.” Reaching that, he mentioned, would require operators, policymakers, and enterprises to work collectively.