The racks are hotter, the bills are bigger and the old playbook is breaking. Artificial intelligence workloads are pushing data center hardware past the limits of air cooling, and the companies running that hardware are running low on something more fundamental than silicon: power and water, according to Scott Patti, vice president of Lenovo's Infrastructure Solutions Group.
Patti oversees the server and storage offerings that enterprises, hyperscalers and a fast-growing crop of AI-first cloud providers rely on. He travels frequently to talk with customers and industry partners.
"I've been in technology for a long time and this is one of the more exciting times of my career," Patti said in a video interview with The Forecast recorded in July at the AMD Advancing AI event in San Francisco.
"Technology's changing so rapidly. With the advancements in AI, it seems like every week or month there's new models which bring new capabilities, which actually impact how we work, how our customers work, how our customers think."
Patti said the demands on customers to keep up have never been more challenging, and the role of an original equipment manufacturer has shifted from shipping boxes to helping buyers navigate a moving target.
How Are Neoclouds Changing the AI Infrastructure Market?
Lenovo's customer base is expanding and growing more demanding. Patti said they span from traditional small and midsize businesses to enterprises to a new category he returns to repeatedly: neoclouds. These are AI-first cloud providers, many of them barely known a few months ago, now deploying immense amounts of capital, power and cooling to offer AI as a service.
"You used to have just a handful of hyperscalers, but now there's all these new emerging Neoclouds, companies you may have never even heard of before, maybe three months ago, six months ago that are now deploying immense amounts of capital, immense amounts of infrastructure, huge power, huge cooling," Patti said.
The neocloud surge reflects a market straining to keep up with AI demand. ABI Research forecasts a potential $250 billion market opportunity for neoclouds by 2030, driven by surging demand for AI-enabling infrastructure and the need for semiconductor makers to diversify their customer bases. North American data center vacancy has fallen to a record-low 1.6%, according to CBRE, pushing organizations toward neoclouds as a pressure valve when they cannot add capacity on site.
Patti said his conversations with neocloud operators now reach well past technology into power, financing and where to physically place a data center.
"A lot of discussion around power, can we get enough power to power all the GPUs that we need? We talk actually a lot about financing and how do they get access to the capital they need in order to deploy the technology," Patti said.
"So while the technology conversation is always there, there's these new elements that are showing up that maybe haven't historically been as top of mind."
Why Is Liquid Cooling Becoming Mainstream in AI Data Centers?
Power and cooling have become a top concern when managing and buliding IT systems. Patti said customers in some municipalities cannot get the power or water they need, forcing strategic decisions about where to build. Lenovo bet on liquid cooling roughly 15 years ago with its Neptune platform, a commitment Patti said he is now grateful for.
"As a technology guy back three or four years ago, you would say, 'Is liquid really going to be something we need?' Well, I'm so thankful that we've stayed committed to it because it's becoming mainstream now," Patti said.
"Customers who had never ever considered liquid, maybe even three years ago are talking to us all the time about: help us get liquid in our data center."
The shift is driven by physics. Newer generations of CPUs paired with GPUs consume 30 to 50% more power, and racks that once drew 17.3 kilowatts now need 34-plus kilowatts with liquid cooling to the chip. Lenovo says its 6th-generation Neptune liquid cooling can deliver energy savings of up to 40% over traditional air cooling. Data centers already account for 2.5% of U.S. electricity consumption and are expected to reach 7.5% by 2030, according to research firm EPRI.
What Is Driving Sovereign AI Concerns Among Enterprises?
As agentic AI systems make decisions and even create data, the question of where that data lives has grown sharper. Patti said sovereignty has long lingered around public cloud but is now top of mind as AI systems train on and generate data.
"Sovereign is a huge one right now, and it's always sort of been there in public cloud around where is my data? Where's my data stored?" Patti said. "Now as agentic is coming along and I'll say systems are making decisions using data, maybe even creating data, where is that data going? How is it being leveraged? What is it being trained on?"
Europe may be the leading edge, Patti said, but the concern is surfacing across geographies. For Patti, the throughline across neoclouds, cooling and sovereignty is the same: customers want to extract value from technology without carrying its complexity. The job of a technology partner, he said, is to absorb that complexity so customers can focus on their business.
"We want to continue to make it easier for our customers," Patti said. "They want to leverage the technology. They don't want to have to be spending time managing new infrastructure, learning new technologies."
Related:
Transcript:
Scott Patti: I'll just say I've been in technology for a long time and this is one of the more exciting times of my career. Technology's changing so rapidly. With the advancements in AI, it seems like every week or month there's new models which bring new capabilities, which actually impact how we work, how our customers work, how our customers think. So I'll just say the dynamics of the industry right now are just, it's a super fun place to be working right now. And the demands on our customers to keep up with the technology, it's actually more challenging than ever for them. So I think that's what's one of the things that we as an OEM in particular, as a technology partner of our customers have to help them with. How do we make it easier for them to keep up with the change? What do they need to be thinking about?What's coming next? Whether it's the next evolution of liquid cooling technology, the next GPU technology, the next CPU technology. We're here at AMD talking about both CPUs and GPUs, right? So I think it's really important right now as technology partners to our customers, helping them through that dynamic changing environment. First and foremost, at the bare most bottom layer, you have to have the infrastructure. So do we have the right infrastructure and hardware for our customers? Then you start moving up the stack. Do we have the right solutions that build on top of that? That's partner ecosystem. Another big thing, I actually spent a bunch of time this week with our services team and our services organization. How do we help a customer deploy, manage? Can we manage for them? Can we offer it as a managed service? So again, ways for our customers to just leverage that technology, leverage the capabilities, not just have to think through all that complexity.
We can take the complexity on so they can just add value to their business. We want to continue to make it easier for our customer. There's new technology involved. If we can allow them to leverage a platform like Nutanix where they can take their historical compute, storage, hyper converge, and then bring AI into that same platform and let them have a consistent operating model is really the key. We do want to make it as simple as possible for our customers because they want to just add value to their business. They don't want to have to be spending time managing new infrastructure, learning new technologies. They want to leverage the technology. So really that's kind of the focus for us is to how do we get the customer to extract value from the technology as quickly as possible? You still have your traditional small, medium business and they're being very, very disruptive by AI, how they leverage it, how do they use it to increase and improve productivity within their business?
The other end of the spectrum, the NeoCloud, you used to have just a handful of hyperscalers, but now there's all these new emerging NeoClouds, companies you may have never even heard of before, maybe three months ago, six months ago that are now deploying immense amounts of capital, immense amounts of infrastructure, huge power, huge cooling. So I'll say that continuum of customer from small, medium business, traditional enterprise, NeoCloud, which is a new term in the market just over the last couple years where we spend a ton of time on in our traditional hyperscalers. So it's really just, I'd say the market is continuing to grow and expand and segment in even a different way than it has historically. A lot of the neo-clouds are focused on obviously providing AI as a service. So when you talk to them, there's so many different dialogues at least I'm having with them that I maybe never had before in my career.
A lot of discussion around power, can we get enough power to power all the GPUs that we need? We talk actually a lot about financing and how do they get access to the capital they need in order to deploy the technology? So while the technology conversation is always there, there's these new elements that are showing up that maybe haven't historically been as top of mind as they are right now. But you have to really think more now about the infrastructure. How much power do I have? How much liquid do I have? How much water access do I have? We have a lot of customers that are now challenged in certain municipalities, not able to get the power they need, not able to get the water they need. So even being strategic about where are they going to place their data centers. And you start seeing this new emerging partners and technologies in things like liquid cooling, CDUs, customers that are actually starting to work and build their own in order to help mitigate some of the supply challenges that they see.
Sovereign is a huge one right now. And it's always sort of been there in public cloud around where is my data? Where's my data stored? Now as Agentic is coming along and I'll say systems are making decisions using data, maybe even creating data, where is that data going? How is it being leveraged? What is it being trained on? So that sovereign notion, which has kind of been in existence for a long time is really becoming top of mind. Certainly certain geographies in the world like in Europe, maybe they're the leading edge of it, but we're seeing it in many, many areas. I mentioned again, power is another top of mind topic and cooling is kind of there with it. So how do we cool in ways that we historically wouldn't? So one of the things I think about as a Lenovo guy is our Neptune liquid cooling, something we've been investing in for about 15 years.
And even I would say as a technology guy back three or four years ago, you would say, is liquid really going to be something we need? Well, I'm so thankful that we've stayed committed to it because it's becoming mainstream now. Customers who had never ever considered liquid, maybe even three years ago are talking to us all the time about help us get liquid in our data center. So that's been an amazing change in the environment just in a couple years.
Ken Kaplan is Editor in Chief for The Forecast by Nutanix. Find him on X @kenekaplan and LinkedIn.
© 2026 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.
Related Articles
