Sean Farney, Vice President, Americas Data Center Strategy, JLL
The explosive growth of AI over the past few months has been a true black swan event. Seemingly overnight, tools like ChatGPT have introduced AI to the masses, creating millions of new users and equally as many questions about its impact on the future of data centers. As major companies put existing data center plans on hold, several important considerations rise to the top as organizations restrategize their investments for the expanding world of AI.
There’s power in thinking differently about site selection.
AI computing requires power. Unprecedented power. At JLL, we’re seeing line-of-site power requirements for new projects in magnitude of up to a gigawatt in single campuses. In top data center markets today like Ashburn, Chicago, Dallas and Phoenix, those sites just don’t exist.
To be ready for an AI future, it’s important to think differently about site selection. New submarkets might hold the answer, and a company’s proximity needs are a key decision factor. For stand-alone, ubiquitous AI tools like ChatGPT, the location of a data center can be more flexible. For banks, retailers and other organizations with proprietary, business-specific AI models to run, proximity to an existing arc and infrastructure may be important. Just remember, planning for power is critical, and the solution might be found in nontraditional places.
Quick takeaway: Expand horizons when searching for sites. Have conversations with utilities in new markets that haven’t traditionally seen data center growth.
Moore’s law is driving data center design.
The future of data center design is directly tied to chip design. Moore’s law, which states that a the speed and capacity of computers will double every other year, is fully in the driver’s seat, and, at some point, even it won’t be able to keep pace with chip advancements. As companies invest billions in new chips to power the future of computing and AI, data centers must be ready to power those chips – and keep them cool, too. Legacy air cooling isn’t an option. Even air-assisted direct chip cooling won’t get the job done. As transistor sizes fall to nanometer scale and below, liquid cooling solutions may be the only viable solution. Those questions are still being answered. But along with adequate power, thermal requirements of chips are a critical consideration for future data center designs.
Quick takeaway: Stay up-to-speed on chip advancements and don’t settle for power and cooling solutions now that will be obsolete five years down the road.
AI is in a tug of war with sustainability goals.
Organizations worldwide are working aggressively toward sustainability and decarbonization goals. Artificial intelligence – and its demand for power and resources – is pulling in the exact opposite direction. Can the two be reconciled? At JLL, this question continues to be an important topic of discussion as we consult with our clients on data center plans. There’s no Magic 8 Ball answer; however, our extensive sustainability work with commercial real estate clients across the globe is starting to illuminate some promising possibilities for making AI more planet friendly.
Quick takeaway: Currently, AI and sustainability goals don’t play well together. But keep an open mind as new technologies and best practices come online that may allow the two to coexist.
At JLL, we’re committed to leading the way in data center site selection, design, build and ongoing operations as we work towards a sustainable future. For more of our perspective on AI and its impact on the commercial real estate industry, read our whitepaper – Artificial Intelligence: Real Estate Revolution? Or Evolution?
Sean Farney is the Vice President for JLL’s Americas Data Center Strategy and works with clients to fully leverage the firm’s capabilities to maximum digital infrastructure value, minimize facility risk and drive data center innovation.
Prior to his current role, Sean was Director of Marketing and Business Development for Kohler’s Data Center business. As Founder and Chief Operating Officer of Ubiquity Critical Environments, a first-mover Edge Data Center Start-Up, he partnered with Schneider Electric and Sears Holdings to design a national network of 50+ sub-1MW facilities located in Tier 2/3/4 MSAs to provide streaming content and services.
He built and operated Microsoft’s 120MW Chicago data center, implementing containerization to drive sustainable operations while deploying 100,000 servers in less than 12 months. Sean has also led Digital Transformation strategic programs at two Fortune 500 companies.
Sean earned a master’s degree in Information Technology from Northwestern University and studied philosophy and education at Miami University.