AI data center infrastructure and operations are now at an inflection point unlike anything this industry has faced before.
Watch John Shultz at PTC 26 on the challenges reshaping AI data center operations—and how Salute is driving the industry to rethink its infrastructure and operations playbook. John Shultz Salute | The Tech Capital
At PTC 26, John Shultz, Chief Product Officer, AI and Chief Learning Officer at Salute, and a leader in direct-to-chip liquid cooling operations and workforce readiness, outlines how the data center market has shifted and why the industry cannot operate the way it did even three years ago.
What’s driving the inflection point?
From 2017 to 2021, the North American data center industry leased roughly 750 MW to 1 GW of capacity per year. Then generative AI arrived. In the first year following the release of ChatGPT in October 2022, the industry leased more than 4 GW of capacity, with 1.5 GW driven by AI alone.
John Shultz describes this significant shift in this way: “Said differently, AI demand in a single year exceeded any full year of data center leasing in history.”
By 2024, total leasing surpassed 6 GW, with over 60 per cent driven by AI workloads. By the end of 2025, the industry was expected to reach 12 GW of annual leasing, with more than 8 GW tied directly to AI.
That scale of growth is unprecedented, but the impact on data center operations is just as big. The introduction of architectures such as direct-to-chip liquid cooling, accelerated by platforms like NVIDIA Grace Blackwell, has pushed power densities beyond what air-cooled data centers were ever designed to support.
This is where John’s perspective becomes critical.
“The last time the industry saw anything comparable was during the internet build-out of the early 2000s, followed by the mobile wave a decade later. But even those shifts did not compress this level of demand, capital exposure, and operational risk into such a short window.”
What makes this moment different is not just the technology but the impact on the safe operations of these facilities. When individual cabinets represent millions of dollars of equipment and liquid is introduced into environments historically designed to avoid it, the safety, uptime, and performance depend on far more than design alone. To protect these enormous investments in AI, it is critical for companies to have an operational model that mitigates the significant risks that are inherent to liquid cooling.
What support is available to upskilling for AI operations?
This is why Salute has focused its AI investment on:
- Expertly trained FM&O teams that are operationally ready for high-density AI environments
- A commissioning discipline and protocols that reflect innovative equipment and infrastructure
- Education and training that prepare operators for unknown risks in environments that have never been built before
In this video and over the coming weeks, we share John’s perspectives on what factors operators underestimate, where the real risks lie in direct-to-chip liquid cooling, and why workforce readiness and safety will determine who succeeds in this next phase of AI infrastructure.
For data center operators, engineers and designers: what challenge is proving harder in reality than it looked on paper?
Call to action.
We are sharing these perspectives to help the industry learn faster. If commissioning, FM&O, or workforce readiness are part of your current challenge, we are actively working in these environments globally.
Get in touch via our AI Hub: https://salute.com/ai-hub/