CoreSite provides high-performance data center colocation, interconnected campuses, and cloud onramps for the hybrid IT infrastructure needs of enterprises in 11 US markets.
Juan, you have been with CoreSite for 14 years. Could you describe your journey in the technology sector, and how the company has evolved in that time?
My career started in the telecom sector in the pre-internet days, focusing on voice traffic and early data traffic. In 2004, I started working in interconnection and colocation services. I joined CoreSite in 2010 at a pivotal moment, just as the company became publicly traded and was positioned for major expansion. This coincided with the rise of the cloud, which transformed how enterprises accessed computing and storage, enabling on-demand services and the first iterations of machine learning.
At CoreSite, I have focused on guiding the company through these industry pivots. Initially, I established relationships with major players like Amazon Web Services, oversaw data center expansions, and optimized returns on invested capital . When American Tower acquired CoreSite in 2021, I stepped into the CEO role. Today, we are preparing for the next evolution: generative AI. This new wave demands even larger, more powerful data center campuses with robust interconnection capabilities to support AI infrastructure clusters.
Why do companies choose colocation for their digital infrastructure?
Historically, businesses operated on-premise data centers, which became costly to retrofit as the infrastructure became outdated over time. We enable businesses to outsource their data center needs in multi-tenant facilities, interconnect and collaborate within digital communities. Whether the deployment is large-scale or smaller, our goal is to support diverse customer needs by providing shared access to critical infrastructure.
Customers typically choose us for three main reasons: interconnection with other parts of the digital supply chain, cost-efficiency compared to running their own data centers, and reliable infrastructure.
The CoreSite Open Cloud Exchange (OCX) has been described as an “easy button” for connectivity. What practical problems does it solve for enterprises, and how?
As the cloud evolved, we became a solution not only for colocation, but to access fast, secure and direct connection to the cloud. With the Open Cloud Exchange®, enterprises can connect to multiple clouds through a single port, supporting increasingly popular multi-cloud strategies. For example, a business may use Amazon for storage, Google for analytics, and Oracle for database management—all seamlessly connected via OCX. This platform reduces complexity and operational costs, offering businesses the flexibility to efficiently manage workloads across multiple providers.
With AI use demanding greater computing density, how is CoreSite adapting its services to meet the needs of AI customers while ensuring energy efficiency?
Enabling AI has long been part of CoreSite’s offerings, supporting machine learning applications in streaming services, social media, and more. Recently, the rise of generative AI, powered by GPUs from companies like NVIDIA, has introduced new demands for power and cooling infrastructure. To address this, we have achieved NVIDIA DGX-Ready certification in key markets such as Los Angeles, Silicon Valley, Northern Virginia and Chicago, enabling clients to deploy high-compute systems in our facilities.
Shared resources in multi-tenant data centers inherently improve efficiency compared to enterprise-owned facilities. Our Power Usage Effectiveness (PUE) is around 1.36, meaning for every kilowatt used by a server, only 36% additional power is needed for cooling. The Uptime Institute states the current industry average for PUE is about 1.5. By contrast, enterprise data centers can have 100% overhead or more. Liquid cooling has been part of our operations for years, leveraging centralized chilled water plants for high-compute workloads. Looking ahead, immersion cooling may become essential for handling densities exceeding 100 kilowatts per rack, as semiconductor advancements drive higher power demands. While this technology is not in high demand yet, our platform is prepared to support it.
How does CoreSite assess the viability of potential clients, particularly those in the fast-moving AI space, to ensure a sustainable business model for both parties?
Not all AI applications are created equal. The training of AI models, which requires massive warehouses of servers consuming immense power, is typically handled by hyperscalers like Amazon, Google, and Microsoft. We do not focus on that segment but instead support smaller deployments from a new wave of companies leasing AI compute on demand. Our primary growth in AI comes from customers who consume AI services for inferencing, spanning industries like financial services, healthcare, higher education, and digital media.
For us, this is business as usual, albeit on a much larger scale. AI is essentially cloud computing on steroids, demanding higher power and cooling densities. According to our latest State of the Data Center report, at least three quarters of respondents are considering moving AI-related workloads from the public cloud, which is increasingly costly, to a colocation data center.
In addition, the report found that cloud interconnection was the No. 1 reason for using colocation for nearly half of the 22 workloads included in the survey. But only 31% of respondents say their current colocation provider offers interconnection to a variety of cloud providers – which shows an unmet customer need.
U.S. data center energy demand is forecasted to double between 2022 and 2035. Meanwhile, 15 states accounted for 80% of data center capacity in 2023. With local grids already seeing electricity supply bottlenecks, what kind of discussions need to be happening, and between which stakeholders, to keep the lights on?
Energy demands are increasing, and collaboration between power companies, local governments, and data center providers is essential to address this. Hyperscalers deal with gigawatts of power, while we focus on megawatts, but the challenges remain significant. Short-term solutions include creative approaches like self-generation and advancements in fuel cell technology. Long-term, innovations such as small nuclear reactors and hydrogen-based systems may become vital.
Data centers are a cornerstone of digital infrastructure, and their role will only expand with AI’s growth. The U.S., as a leader in AI technology, must prioritize maintaining this position. For our part, we are ensuring sufficient capacity in land and power to continue serving customers, while broader solutions to these energy challenges will require multi-party commitment.
Are there any upcoming expansion plans beyond CoreSite’s current 11 U.S. markets?
We are actively expanding in Denver, Santa Clara, Los Angeles, New York, and Virginia, focusing on increasing capacity in existing markets where demand is outpacing supply. We are also exploring new regions where cloud and AI infrastructure are emerging, and which will require digital hubs to enable connectivity with these utilities. In the longer term, we see opportunities in edge data centers, particularly as advancements in 6G drive demand for compute resources closer to both wireline and wireless end users. Our combination with American Tower positions us uniquely to deliver end-to-end solutions as wireless and wireline networks converge at the edge, and to support the next wave of latency-sensitive applications and distributed compute needs.
How does CoreSite respond to risks posed by technological disruption or new competitors in the AI infrastructure space?
With any major technological disruption, there is always a risk of disintermediation. The capital investment in AI infrastructure, driven by hyperscalers, is unprecedented, far surpassing even the dot-com boom. However, unlike the dot-com era, these investments are backed by major players like Microsoft, Meta, or Amazon, making them far less speculative in nature. The longer term challenge for hyperscale developers lies in commoditization and returns on capital investment compression as supply and demand reaches equilibrium.
We are taking a measured approach, avoiding reactive investments in oversized facilities. Instead, we focus on evolving our portfolio to meet new use cases, such as higher power densities and increased connectivity demands. Our strategy prioritizes long-term scalability while adapting to the staggering growth in data generation and AI adoption, ensuring we remain a critical enabler for enterprises navigating this new era.
