Podcast Summary: "The State of Play of Data Center Development"

Podcast Information:

• Title: Catalyst with Shayle Kann
• Host/Author: Latitude Media
• Episode: The State of Play of Data Center Development
• Release Date: June 12, 2025

Introduction

In this episode of Catalyst, hosted by Shea Khan, Shayle Kann delves into the burgeoning world of data center development with a special guest, Chris Sharp, the Chief Technology Officer (CTO) of Digital Realty. With over 20 years of experience in developing, owning, and operating co-located data centers globally, Chris provides valuable insights into the current landscape and future trends of data center infrastructure, especially in the context of the increasing demand driven by artificial intelligence (AI) and other high-performance computing applications.

Geographic Dispersion of Data Centers

Shea Khan initiates the discussion by addressing the geographic dispersion of data centers in the United States. Historically concentrated in regions like Northern Virginia, the data center landscape has seen a significant spread due to the surge in AI and hyperscale data center planning.

Key Points:

• Workload Distribution: Chris Sharp emphasizes the distinction between training and inference workloads. While training large AI models initially drove broader regional deployment, current trends show that inference workloads are prompting more region-specific growth.

  “AI is an 'and' not an 'or' to cloud.” (05:12)

• Availability Zones: The concept of Availability Zones (AZs) remains pivotal, particularly in major markets like Northern Virginia, which continues to be a critical hub for AI growth due to its robust infrastructure and proximity to major city centers.

Power Constraints and Availability Zones

A significant portion of the conversation revolves around power availability and its impact on data center development within key regions.

Key Points:

• Power Saturation: Chris highlights that regions like Northern Virginia are approaching power saturation, with vacancy rates as low as 0.5%, indicating a near-maximum capacity for new data centers.

  “Some of the most recent stats, it has 0.5% vacancy rate, which is phenomenal.” (12:23)

• Utility Coordination: Effective collaboration with utility operators is crucial. Digital Realty invests in master planning and works closely with utilities to ensure they can meet the substantial power requirements of new data centers.

• Infrastructure Constraints: Long lead times for transformers and switchgear (over 50 weeks) present significant challenges, extending the timeline for data center development beyond the typical two-year window.

  “Transformer lead times are 50 plus weeks right now.” (15:51)

Timeline of Data Center Development

Shea Khan probes into the timelines involved in developing new data centers, contrasting current durations with historical benchmarks.

Key Points:

• Standard Timeline: Typically, the process from concept to operational data center spans approximately two years.

  “It takes about two years from concept to delivery to build out what I would say a versatile data center.” (14:39)

• Current Challenges: Due to power and equipment bottlenecks, timelines are elongating, sometimes surpassing four years, compared to the usual two.

  “It's elongating even beyond the 24 months that it would take us to pull that together.” (14:39)

Scale of Data Centers and Workload Requirements

The discussion shifts to the scale of data centers and how different workloads influence data center design and capacity.

Key Points:

• Diverse Needs: Not all data centers require massive scales. While some AI workloads necessitate gigawatt-scale centers, others operate efficiently with smaller capacities (e.g., 5-10 megawatts).

  “Not everybody needs 100 megawatt data centers.” (17:58)

• Workload Specificity: High-density GPU infrastructures are in demand for training, whereas inference workloads benefit from smaller, more flexible deployments embedded within existing data center portfolios.

  “Inferences can come in like 5-ish megawatt blocks.” (19:17)

Bridge Power and Utility Relations

A critical challenge in data center development is the interconnection to the power grid. Chris Sharp discusses strategies like bridge power to mitigate delays in grid interconnections.

Key Points:

• Bridge Power Usage: Although rare, some data centers employ bridge power solutions—using generators or on-site power sources temporarily until grid connections are established.

  “Bridge power is one tool among many that we're always looking at.” (26:32)

• Utility Collaboration: More common than bridge power is the proactive collaboration with utilities to negotiate interruptible tariffs or other agreements that facilitate faster power integration.

  “It's a negotiated sort of deal with the utility...” (29:14)

• Avoiding Grid Overreach: Digital Realty ensures that they are not contributing to overbuilding by aligning closely with utility operators and committing to long-term utilization of the infrastructure.

  “We form long term relationships with the utility operator...” (24:22)

Future of Data Center Infrastructure

Towards the end of the episode, Shea Khan and Chris Sharp explore the future advancements and innovations in data center technology.

Key Points:

• Efficiency Improvements: Innovations in liquid cooling are enhancing efficiency, as liquid can be 800 times denser than air, allowing for more compact and efficient cooling solutions.

  “Liquid's 800 times denser than air, so you can get more efficient.” (33:55)

• Advanced Hardware Designs: Collaboration with technology partners like NVIDIA is driving the development of next-generation AI hardware, improving performance and energy efficiency.

  “The ability of the token production against the watts associated with that is going to be phenomenal.” (33:55)

• Sustainable Practices: Emphasizing stewardship of power resources, Chris underscores the importance of sustainable practices in powering data centers to support both environmental goals and operational efficiency.

  “We want to be good stewards of the power we take from the grid.” (33:55)

Conclusion

The episode concludes with an optimistic outlook on the role of data centers in advancing AI and supporting climate-tech initiatives. Chris Sharp expresses enthusiasm for projects like Gephion in Copenhagen, which leverage large-scale data centers to drive significant advancements in fields such as pharmaceuticals.

“The amount of pharmaceutical work associated with that one deployment is just what it's going to be able to do for humanity is very exciting to me.” (35:18)

Shea Khan thanks Chris Sharp for his insights, wrapping up a comprehensive discussion on the current state and future trajectory of data center development.

Notable Quotes:

• “AI is an 'and' not an 'or' to cloud.” – Chris Sharp (05:12)
• “Some of the most recent stats, it has 0.5% vacancy rate, which is phenomenal.” – Chris Sharp (12:23)
• “Transformer lead times are 50 plus weeks right now.” – Chris Sharp (15:51)
• “It takes about two years from concept to delivery to build out what I would say a versatile data center.” – Chris Sharp (14:39)
• “Not everybody needs 100 megawatt data centers.” – Chris Sharp (17:58)
• “Inferences can come in like 5-ish megawatt blocks.” – Chris Sharp (19:17)
• “Bridge power is one tool among many that we're always looking at.” – Chris Sharp (26:32)
• “We form long term relationships with the utility operator...” – Chris Sharp (24:22)
• “Liquid's 800 times denser than air, so you can get more efficient.” – Chris Sharp (33:55)
• “The amount of pharmaceutical work associated with that one deployment is just what it's going to be able to do for humanity is very exciting to me.” – Chris Sharp (35:18)

Additional Information:

For more insights and episodes on climate tech and energy transitions, visit Latitude Media.