Flexential CEO Chris Downie outlined a brand new playbook for knowledge middle 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, wherein each he and Klitschko had been smiling and good-naturedly flexing for the digicam. Inside quarter-hour, the good friend despatched again a short video, seemingly of that very same second: The lads 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 may 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 information middle business, 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 Information Middle Frontier Tendencies Summit. “And in the event that they may rewrite my face and the scene of which I’m in, in lower than quarter-hour, what’s it going to do to the business?
“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 plenty of the up-and-down cycles the information middle business has already weathered — he thinks this time is basically completely different. “From my perspective, this isn’t non permanent. And that is the second that’s going to redefine our function as operators, as infrastructure enablers within the business,” he stated.
Information middle operators have kind of adopted the identical playbook in previous cycles, he stated, tweaking and adapting based mostly on classes realized 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 stated, including: “As we glance ahead, we have to respect that all the pieces’s completely different. And doing issues enterprise as typical, the identical previous means, isn’t going to use.”
So what are the brand new guidelines for knowledge middle operators? Downie outlined the foremost business touchstones which can be being utterly reshaped.
–Energy wants. That is maybe the most evident crunch throughout the information middle business. Downie cited numbers from Reuters that indicated the typical load from AI and knowledge facilities this 12 months is projected to hit 488 gigawatts — the very best ever. Provide shouldn’t be protecting tempo, with annual demand outstripping utilities’ capability to supply. The hole between what the grid can present and what knowledge facilities want may exceed 20 gigawatts by 2030, he warned, which creates cascading challenges for DC operators.
These cascading results embrace utilities shifting prices to knowledge middle operators in ways in which they haven’t beforehand, and demanding up-front funding. “That was not within the previous playbook,” he stated. “So when you consider constructing substations or transmission strains, you now have to consider doubtlessly a whole lot of hundreds of thousands of {dollars}, upfront of deployment.” That was 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 known as “hyperspeculators”, are snapping up properties that would doubtlessly function knowledge middle 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 should advance capital, to get land nicely upfront of after we in the end wanted to prior to now,” he stated.
–Construct timelines preserve stretching additional and additional out. Within the previous 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 circumstances, the wait time for energy grid connections are exceeding seven years. If transmission initiatives are concerned, the wait might be even longer. That stage of delay, Downie stated, could also be “going past the horizons which can be doubtlessly set by the traders which can be fueling the capital to drive our infrastructure.”
–Provide chains are stretched. Connection to energy isn’t the one bottleneck. Crucial tools like distribution transformers can face lead occasions of as much as 5 years, based on Downie. “That delay will affect even the best-prepared operators,” he stated. “And we’re not speaking about simply transformers. You’ve obtained swap gear, mills, cooling techniques … bodily plant has develop into a strategic problem. So, the brand new playbook will drive us to suppose extra dynamically about planning, procurement, and constructing than maybe we’ve ever skilled prior to now.”
–Constraints on elasticity. Prior to now, knowledge middle operators may assume that there was some predictable elasticity in provide, the availability chain, compute, expertise and the essential uncooked supplies wanted for constructing and working knowledge 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 stated.
–Infrastructure going from passive to pivotal. Digital infrastructure, Downie stated, “has moved from the boiler room to the entrance strains.” That’s a brand new place for the business. “For 17 of my 20 years within the enterprise, knowledge 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. Which means the information middle operators are transferring from being basically “order takers” to “innovation enablers.”
–Capital isn’t equally distributed, and prices are excessive. The widespread assumption is that the billions of {dollars} pouring into the information middle market will deal with the calls for of hyperscalers. However the place does that go away the smaller enterprise clients who’ve stuffed knowledge facilities for the previous twenty years? Their wants aren’t going away, and so they need to undertake AI instruments as nicely — however now there may be extremely intense competitors for knowledge middle assets. Downie stated that he has needed to resolve whether or not to allocate knowledge middle 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 fascinated by the way it thinks about elasticity in its personal new playbook,” Downie suggested. Which means fascinated by procurement timelines and being real looking about prices. Within the new playbook, Downie sid, “All the things is dear. 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 stated, 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, neighborhood relations, touchstone expertise — lots 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 12 months in the past. “Issues are positively accelerating. … There’s excessive confidence that AI roadmaps have to 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 shall be judged by their capability to make onerous trade-offs. He acknowledged having to prioritize capability for one buyer over one other in latest months, a call he by no means beforehand contemplated. “Shortage goes to separate the leaders from the loud,” he stated.
Regardless of the challenges, Downie remained optimistic concerning the business’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 world scale.” Attaining that, he stated, would require operators, policymakers, and enterprises to work collectively.
