At the sidelines of Canalys’ annual conference in Bali, Schneider Electric shared about the cooling technologies initiatives they have in anticipation of rising compute demand. It is good to be prepared, but will this demand actually materialize, or is the industry getting ahead of itself?
According to Paul Tyrer, Schneider Electric Global VP of Channel IT Ecosystem, the company’s cooling and power solutions can be found in big hyperscale players, large co-location data centers, right through to enterprise and edge environments.
The relevance of their solutions couldn’t be more critical than it is now.
With artificial intelligence driving unprecedented demand for computing power, data centers are facing new challenges in power density and cooling requirements. Schneider Electric estimates that over the next five years, an additional 150 gigawatts of data center capacity will come online, with 60% dedicated to AI workloads including large language models and inferencing tasks.
Notably, this growth in power demand is outpacing current demand trends, highlighting just how much this surge in AI adoption is pushing traditional cooling solutions to their limits.
What does the channel ecosystem say about AI?
Canalys chief analyst Alastair Edwards shared survey results of gradually more channel partners accepting AI as a business opportunity. “For AI, fast forward to 2024 you have got about 34% of surveyed partners saying that basically it is a tremendous opportunity; it’s still a minority but a big increase.”
Alastair resonated with the conference’s keynote presentation by Steve Brazier who had observed that there is the short-term opportunity that most partners are focused on, but a long-term anticipation for real adoption of AI by enterprises to pick up.
“From an enterprise perspective, we really feel the transformation opportunity that this represents is going to transform the market, and by association, the sort of revenue and growth of profit for partners.
“So, everyone who’s really not thinking about this and preparing their business is going to be left behind.”
While Canalys has forecasted partner services revenue associated or wrapped around AI to grow to an estimated USD160 billion by 2028, Alastair cautioned that there would be unequal distribution of revenue and opportunity as only some partners will get to the forefront of this and some will gain a lot of benefit in the short term.
Notably Alastair noticed a lot of AI activity amongst:
- Large SIs – their clients want to define business transformation and big management consultancies like Bain and Boston Consulting Group are building Ai practices to define this as well as plan how to translate it into technology opportunity
- Partner ISVs or independent software vendors are building software with AI features that solve particular use cases
- Channel partners are building data practices as it is the fuel for AI and clients want to ensure that volume and quality of data are adequate
The chief analyst noticed a number of proof -of-concepts that try to work out AI and how to use it. “The real challenge is getting that to production and getting that to scale…. Most times the directive will come from the top and it gets pushed down to the IT teams and execution teams, and that’s where it grinds to a halt.”
Besides cost, barriers to a production rollout throughout an organization tends to be access and quality data.
Alastair admitted there is a real opportunity now for channel partners to re-architect their IT environments, create new business models and build next-generation infrastructure to support new AI workloads, that over time will drive a wave of investment.
Schneider Electric’s opportunity: Power and cooling
While conventional data centers typically handle 10-20 kilowatts per rack, AI workloads are demanding between 40-100 kilowatts, with some high-performance computing systems reaching up to 400 kilowatts per rack.
To address these challenges, Schneider Electric recently announced its acquisition of MotivAir, a specialist in liquid cooling technologies. Expected to conclude by February 2025, this strategic move positions Schneider Electric to offer comprehensive cooling solutions from “grid to chip and chip to chill” – an end-to-end approach that manages power from the electrical grid to the processor, and handles heat dissipation from the chip to the cooling system.
Paul described that the acquisition “… basically gives us the full backbone from the mechanical, the cooling, and the electrical, as well”
The urgency of these cooling innovations is underscored by the massive investment flowing into data center operations. In 2023 alone, USD36billion was invested in dafa centers globally. In the first five months of 2024, USD22 billion was invested globally in data centers. This acceleration reflects the industry’s recognition that traditional infrastructure must evolve to support AI’s computational demands.
Paul observed a lot of GPU implementations are going into training large language models and hyperscale environments.
“We firmly believe that as inferencing really starts to become more mainstream, enterprises getting onboard will become the domain where the channel has got a very significant role to play. And I think in many ways that we don’t even realize yet.”
