Boston Data Center Summit: Chris Coleman on AI & HPC — Which is Driving More Demand?
by Josh Anderson
BOSTON, MA — Chris Coleman currently leads Digital Realty’s Enterprise Strategic Accounts group across North America. His team focuses on worldwide data center needs for technology, software, financial, and healthcare companies. Chris has over 20 years of management and sales experience in the data center industry. He spent thirteen years at American Power Conversion (APC) where he held a variety of positions, including Director of Government Sales and Director of Enterprise Business in Europe. Before Digital Realty, Chris was Vice President of New England for SunGard,focusing on disaster recovery, managed services, and consulting services. At CapRE’s 2018 Boston and New England Data Center Summit, Chris participated in a Q&A titled “Boston Data Center Market Overview & Analysis” moderated by Kevin Normandeau, Publisher at Data Center Frontier. Below, we highlight the first few topics addressed by Coleman and Normandeau.
Kevin Normandeau, Publisher, Data Center Frontier: So Chris, there was a lot of discussion about AI in HPC. These particular technologies are attractive to universities and biotech firms. Tell us a bit about whether these new applications are boosting demand. Which ones are driving more demand than others?
Chris Coleman, Vice President – Enterprise Sector, Digital Realty: I think of HPC and AI from a technology perspective. The data center perspective can be very similar. I look at HPC as kind of academic and research oriented. I look at AI applications are more implementing HPC into business processes and driving those processes. So we have the MA HPC Center out in Western Massachusetts, it was a consortium of universities here in Massachusetts and they’re doing climate studies, forest mapping, all sorts of, I think, academic or pure research type of work. Whereas, for us, and our data center demands on the AI side, it is really exciting.
We’re working with a couple of clients right now – they run their corporate data centers in a couple of our sites and they may have 800,000 square feet of data center. And one example is a retail chain, who is trying to figure out through AI how they can communicate with their customers more effectively to get them to come into their site during off-peak hours, and they’re running an AI pilot right now that, from a data center perspective, it’s a 400 KW footprint that they’ve put into test out this pilot. And it’s communicating to a limited amount of the customers that go into their 10,000 stores. If the pilot is successful, they’re going to need 4 megawatts of data center capacity. So we have an enterprise client who was, probably using less than a megawatt and going down, who now could be looking at a 4X growth in data center demand as they roll out an AI application.
Normandeau: So it sounds like AI is driving more demand than HPC. Is that what I’m hearing?
Chris Coleman: Yeah, I think that it’s, to me, a lot more widespread in the applications.
Normandeau: I think that in terms of what’s interesting about HPC and AI, is that HPC tends to be thought of more of an academic or scientific [venture] that does parallel processing, and AI seems to be more popular with the enterprise group, but it’s still a lot of parallel processing. So I would think that from a data center perspective the demands might be similar between AI and HPC, it’s just that one client might call it something slightly different than the other.
Coleman: You’re right.
For more from Coleman, check out previous CapRE Insider Reports: