While Covid-19 may now feel like a thing of the past, its impact on the data center industry has been long-lasting. Global lockdowns severely disrupted supply chains, posing serious challenges for an industry where infrastructure is mission-critical.

Throughout the early 2020s, data center operators faced long delays, often contending with multi-year backlogs from major manufacturers for the equipment they needed to stay operational. In this episode of DCD>Talks, Dennis Strieter, co-founder and chief revenue officer of Hyper Solutions, delivers a masterclass on navigating supply chain challenges and meeting the evolving power demands of AI-driven data centers.

A long time ago in a galaxy far, far away…

Well, 27 years ago, to be exact – Strieter and his team began manufacturing mission-critical equipment for the data center marketplace. With products like power distribution units (PDUs), remote power panels (RPPs), static transfer switches (STS), and switchboards, Hyper Solutions launched with a mission to build infrastructure that lasts. Strieter and his partners set out to rescue the industry from supply chain chaos:

“We created a new manufacturing model – a distributed network of plants that isolates supply chain problems. Instead of relying on one or two massive factories, like most traditional manufacturers, we operate nine facilities spread across the US.”

Making the jump to lightspeed

Over the past quarter of a century, Strieter has witnessed a dramatic evolution in data center power densities, growing from around one kilowatt per rack to, by 2018, advanced colocation facilities supporting 20 to 30 kilowatts per rack. At that time, this was sufficient to meet the needs of the traditional enterprise IT stack and the emerging cloud colocation market.

Then Google introduced the transformer model, and the AI acceleration race began. With it came massive demand for compute, driven largely by GPUs like Nvidia’s:

“We all know the H100. When it first hit the market, deploying a full rack meant drawing around 72 kilowatts per rack. That’s two to three times beyond what most facilities were built for. And Nvidia didn’t wait around – soon after, the H200 launched, pushing densities to 132 kilowatts per rack.”

So when we talk about “future-proofing” data centers, the real question becomes how to plan for a future that’s evolving faster than infrastructure can adapt.

“None of us has a crystal ball, but we can draw some reasonable inferences from the direction things are heading,” Strieter adds.

Specializing in supply chains

With the AI era now in full swing, Strieter is no stranger to clients requesting new infrastructure with high-capacity output foot feeds to power next-generation AI stacks. The Hyper solution? Modular engineering:

“Our engineering team breaks each piece of power distribution equipment into logical modules. Each section is designed as a standalone unit within the overall product frame. That way, we can quickly swap or redesign just one module – rather than starting from scratch – allowing us to move fast.”

As a private manufacturer, Hyper Solutions focuses on delivering the innovative features its clients need – and avoiding those they don't – to help them achieve speed, performance, and financial goals.

From solution development through manufacturing execution, Hyper Solutions relies on its digital system, HyperSpace, to scale intelligent, data-driven production.

“At our headquarters in Richmond, Virginia, you can look at the big board and watch in real time as a new product begins its life at any one of our factories,” says Strieter, adding:

“Every nut, bolt, and torque specification is digitally recorded by the assembler and instantly visible to us. As each product moves through manufacturing, we can track its progress down to the smallest detail – and we’ve extended that visibility to our clients, too.”

Cooling coming soon

AI-specific data centers face a major challenge in managing power, cooling, and sustainability – particularly how to bring liquid into the data center, integrate it with sensitive high-tech equipment, and still maintain an environmentally positive footprint. Strieter contextualizes:

“We've determined that water must be delivered directly to the chip – that's the requirement. You can't afford any interruptions. That means your mechanical engineers now have the critical task of designing highly available, redundant systems to ensure water continuously flows to those chips. If it stops, the system has to shut down.”

Looking ahead, Strieter introduces Hyper Solutions’ next-generation venture – repurposing water used in cooling, rather than discarding it. He exclaims that the goal is to turn this necessity into a strategic advantage:

“At Hyper, we view this challenge as an opportunity. By rethinking how we use cooling water within our own equipment, we aim to miniaturize our footprint – freeing up more white space for IT racks and reducing reliance on bulky PDUs and RPPs.”

No one’s ever really gone

According to the Uptime Institute, more than 50 percent of data center facilities engineers will retire by the end of 2025. At the same time, massive hundred-megawatt data centers are racing to meet soaring customer demand, creating a major logistical glitch.

Enter HyperSense AI, Hyper Solutions’ AI and machine learning platform, designed to bridge the growing workforce gap. Strieter explains:

“As a manufacturer, we can begin to integrate and extend the human senses – sight, sound, touch, taste, and smell – into our power distribution equipment through advanced sensor technology.”

With continuous thermal metering, HyperSense AI enables clients to remotely monitor the temperature of every connection, eliminating the need for physical inspection and exposure to hazardous voltages. By correlating thermal data with power output, operators can predict failure points, optimize usage, and receive real-time alerts, including warnings when integrated sensors detect a loosened bolt.

Planetary-scale AI access

As AI becomes part of the fabric of daily life, Strieter challenges the industry to think beyond just who has access today:

“Imagine Universal Basic AI: like Universal Basic Income, but guaranteeing every person on the planet access to a baseline AI assistant. That means extending AI access to developing nations and underserved communities.”

To the builders and operators of this new era: design with inclusion in mind. The AI revolution must lift everyone, or it won’t take us nearly as far.

Watch the full DCD>Talks episode with Dennis Strieter of Hyper Solutions here.