Data Center Seminar Series

Topics will include the challenges of keeping data centers cool, the impacts of data centers on electricity rates, community perceptions of data center development in PA, and more. We hope the series enables many new interdisciplinary research collaborations and research-extension partnerships.

Tuesday, February 24, 2026

The Challenges and Opportunities in Keeping Data Centers Cool

Wangda Zuo - Professor of Architectural Engineering, Penn State

Time: 12:00 - 1:00 pm

Location: Food Science 342

Abstract: Driven by the rapid growth of artificial intelligence, data centers are projected to consume up to 12% of U.S. electricity by 2028, up from approximately 4% in 2023. Cooling alone accounts for roughly 40% of total data center energy consumption, making it a critical challenge for both operators and the power grid. In parallel, the growing scale of data center deployment has prompted increased attention to their interactions with local energy systems and water resources. This seminar will first review the current landscape and emerging trends of the data center industry. It will then introduce major data center cooling technologies and discuss the technical and operational challenges associated with maintaining thermal reliability at scale. A real-world case study of a data center in Massachusetts will be presented to illustrate these challenges in practice, along with insights drawn from ongoing academic–industry partnerships with data center facilities in Pennsylvania. Finally, the seminar will highlight recent research advances aimed at improving cooling efficiency and discuss how these advances create new opportunities for more sustainable and resilient data center operations.

Time stamps:

• 00:27 What is a data center?
• 01:02 Trends in data center electricity use
• 01:47 Trends in data center water use
• 02:43 Existing and proposed data centers in PA
• 03:58 Cooling accounts for ~40% of total energy consumption in data centers
• 04:30 Types of data center cooling systems
• 06:21 Challenges with data center cooling
• 09:50 Opportunities for making data center cooling more efficient
• 10:26 Developing software for energy-efficient data center cooling
• 13:37 Case study with a Massachusetts data center
• 16:24 Energy saving opportunity #1 (replace degraded cooling coils)
• 17:17 Energy saving opportunity #2  (use smart control to take advantage of cold outdoor air)
• 19:41 Energy saving opportunity #3 (eliminate simultaneous heating and cooling)
• 20:18 Energy saving opportunity #4 (optimize floor air temperature setpoint)
• 21:26 Optimization strategies and potential savings
• 22:52 Using a digital twin to test different scenarios
• 23:39 Energy savings of 53-74%
• 25:55 Optimizing for different climates
• 27:42 Using physics-informed AI to optimize energy and operation costs
• 30:30 Using underground (geothermal) cooling
• 31:00 High-temperature liquid cooling
• 33:53 Heating swimming pools, greenhouses, homes
• 35:06 Honeybee-inspired electric grid control
• 35:54 Summary and conclusions

 Q&A

• 36:54 How does energy usage correlate to water usage?
• 37:29 How would an HVAC technician interact with these smart control systems?
• 38:56 Opportunities for mixed/augmented reality, especially for operation training
• 40:23 Is the heat from the chillers being dispersed into the air? Where does that heat go?
• 41:24 Why don't we see more geothermal cooling being used? Is it because the geothermal loops would need to be extremely large?
• 42:39 How much does the quality of the water change when it goes through a data center cooling system, and do data centers discharge a lot of water to the environment?
• 44:43 What can be done to address heat pollution from data center wastewater?
• 45:33 Given that 53 new data centers are being proposed in PA, how significant is this for environmental harm? Should we try to reduce our personal usage of AI? And then, where internationally do we see the most data center construction happening?
• 48:37 Can we expect to see small modular reactors in PA where new data centers are being proposed?
• 50:30 How feasible is Jensen Huang's solution of new chips that don't need a chiller (slide 32)? Since dry cooling typically consumes more energy, how do you reconcile that tradeoff?
• 51:57 How do you see the role of digital twins and CFD-based control evolving from design tools into real-time operational decision systems for data center cooling
• 53:12 As new data centers are being built and as technology keeps advancing, do you foresee the need to make frequent retrofits to these smart cooling systems? Is there a way to design cooling systems to anticipate and avoid the need for retrofits so that the adoption of these smart systems becomes more attractive?
• 54:57 How do smart control systems optimize cooling when data centers are built with a much larger footprint than is actually needed (due to the increase in computing power and racks becoming smaller and more dense)?

Monday, March 23, 2026

Examining Public Concerns about Pennsylvania Data Centers

The AI Data Centers and Infrastructures Learning Lab (Natalie Rae, Christina Reeves, and Jusil Lee)

Time: 1:00 - 2:00 pm

Location: Food Science 342 and Zoom

Abstract: The proliferation of AI technologies parallels the accelerated development of data centers in Pennsylvania, where communities are navigating the impacts of AI-infrastructures as lived realities, not abstracted debates. In this talk, we share preliminary findings from an ongoing case study examining how residents engage with and articulate civic concerns around new data center proposals for AI technologies.

Notes: Graduate students Christina Reeves and Jusil Lee delivered the main presentation, followed by the Principal Investigator, Natalie Rae (Assistant Professor of Learning Design & Technology), who fielded questions from both online and in-person attendees.

Time stamps:

• 00:00 Background
• 01:55 Tracking data center proposals across PA
• 03:52 Data center siting, zoning, and the PA Sunshine Act
• 05:05 Conceptual framework and the research question
• 07:56 Case study context (Lancaster City council meetings re: data center development)
• 12:04 Concern #1 Strain on municipally managed water, power, and waste management utilities
• 15:27 Concern #2 Lack of transparency and accountability in the municipal process
• 18:02 Concern #3 Air pollution and quality of life
• 23:19 Concern #4 Benefits and the harms of data centers shared unequally
• 25:04 Synthesis and conclusions

Q&A

• 30:27 What proportion of the data centers in PA are supporting AI versus other types of cloud storage services?
• 33:32 Did the public commenters in your study make a distinction between data centers for AI versus cloud storage?
• 35:00 Have you considered applying an "energy justice" strategy or methodological approach in future work?
• 36:50 Are there any municipal requirements in Lancaster or other PA cities to disclose water and power requirements?
• 38:22 What audiences do you have in mind for sharing these findings, and how would you like these findings to be taken up?
• 40:54 Has there been any connection with county planning commissions to compare and contrast how each county is approaching their data center planning?
• 42:30 Given the potential imbalances between the financial benefits of data centers and the environmental and public health costs, what types of measures can communities take at the legal level or policy level to protect themselves? 

Monday, April 6, 2026

Managing Data Center Impacts to Communities: Efforts to Protect Residents by PUCs, States, and Local Governments

Hannah Wiseman and Michael Helbing

Time: 11:30 am - 12:30 pm

Location: Food Science 342

Abstract: As data center proposals proliferate across the Commonwealth and the nation, state and local policymakers are weighing how best to respond. Decisions by federal and state regulators, local governments, and utilities themselves will shape how energy costs and other potential impacts are managed. This session will provide an overview of how states and municipalities are addressing data center proposals, highlighting examples of statutes and ordinances enacted to guide development and protect community interests.

We will focus particular attention on efforts to protect residential customers from bearing increased costs attributed to data center growth. In some parts of the United States, electricity demand is burgeoning, and nearly all of this demand is driven by new data centers. In the PJM region—the network of transmission lines that covers Pennsylvania, the Mid-Atlantic, and parts of the Midwest—30 gigawatts of the 32 gigawatts of new generation capacity projected through 2030 is for data centers. In Georgia, approximately 80 percent of Georgia Power’s new load is caused by data center growth. Most data centers are seeking services from electric utilities that serve all other customers in a region, including residential, commercial, and industrial customers. And in some places, non-data center customers are paying for the costs that new data centers impose on the grid as they demand more generation, transmission, and distribution. Many states and individual utilities are addressing this cost-shifting through updated “large load” rates and policies, but it is not yet clear how effective these rates will be. This presentation will discuss the impact of data center growth on electricity rates and assess the tools available to ensure that data centers pay for the costs that they impose on the grid. The presentation will also briefly examine the beneficial services that data centers can provide to the grid—namely, demand response during periods of peak demand—and how rates can encapsulate these benefits.

Slide deck: Managing Data Center Impacts to Communities

Timestamps and links:

• 00:00 Introductory comments
• 02:38 Municipal action on data centers
• 05:42 Approaches to data center siting
• 07:18 Three mechanisms for approval of use (by right, conditional use, special exception)
• 08:51 Minimum lot size restrictions
• 09:26 Setbacks
• 11:11 Building height restrictions
• 11:35 Requirement to submit a study or plan
• 12:53 Miscellaneous provisions (color, underground utility lines, security fencing)
• 13:35 Existing PA laws/permits that apply to data center development (Chapter 102 and 105 permits, NPDES permits, air quality permits, water withdrawal approvals)
• 15:40 Proposed PA House bills pertaining to data centers
• 18:29 Proposed PA Sentate bills pertaining to data centers
• 19:53 PA Public Utility Commission activities
• 20:52 Data center definitions and grandfathering
• 22:12 Sales tax exemptions
• 23:00 State policies and utility settlements
• 24:40 Electricity load growth projections
  • 2024 United States Data Center Energy Usage Report
  • Letter to PJM stakeholders (August 8, 2025)
  • PJM map
  • 2024 State of the Market Report for PJM - Section 12 Generation and Transmission Planning
  • Electricity generation capacity for California
• 26:38 Overview of state regulations
  • Ensuring a National Policy Framework for AI (Executive Order 14365, December 11, 2025)
• 28:08 State permitting provisions for data centers
  • Data Center Regulation Tracker
• 32:00 State legislation regarding increased utility costs
  • The study that Hannah mentions at 34:10 is Rethinking Load Growth: Assessing the Potential for Integration of Large Flexible Loads in US Power Systems
• 34:32 The problem of overbuilding electricity generation capacity, recommendations to avoid overbuilding include:
  • Texas Senate Bill 6 (large loads must disclose whether they are "pursuing a substantially similar request for electric service")
  • Recommendation from American Electric Power (AEP) to the Public Utilities Commission of Ohio (October 23, 2024)
  • PJM Critical Issue Fast Path Stage 3 Proposal
• 36:01 The problem of stranded costs
• 36:47 The problem of cost shifting (residents paying for data centers' infrastructure costs through higher electricity rates)
  • Customers in 7 PJM states paid $4.4B for data center transmission in 2024
  • Loophole Costs Customers Over $4 Billion to Connect Data Centers to Power Grid
  • Extracting Profits from the Public: How Utility Ratepayers Are Paying for Big Tech’s Power
  • Georgia Public Service Commission Approves Rule to Allow New Power Usage Terms for Data Centers
  • Texas Senate Bill 6 ("large loads must pay contribution in aid of construction")
• 38:35 Incentives to reduce electricity consumption
  • South Dakota Electric Rate Schedule
  • Black Hills Energy Schedule of Rates for Electric Service
• 40:05 Are data centers harmful to agriculture?
• 43:20 Given the enormous scale of data center and AI growth, do we need policies at higher levels than just municipalities and states? (Maybe, but the risks and consequences of regulatory capture could be higher)
• 46:24 Are there regulations or safeguards related to water quality impacts of data centers?
• 49:00 Have you come across examples where developers have offered communities incentives like money for parks, trails, etc. to allow the development of a new data center?
• 51:17 Have there been any legal cases challenging the secrecy associated with non-disclosure agreements that data center companies have with local governments when they're proposing new data center development?

Wednesday, April 22, 2026

Integration of AI Data Centers to Bulk Power Grids: Challenges and Opportunities

Nilanjan Ray Chaudhuri

Time: 11:00 am - 12:00 pm

Location: Food Science 342

Abstract: The rapid proliferation of AI-driven data centers (AI DCs) presents unprecedented challenges for their integration into bulk power grids. These facilities exhibit highly volatile, computation‑driven power demand profiles, pushing the limits of traditional grid planning, operations, and reliability frameworks. In this talk, I will outline the key technical and operational challenges posed by large‑scale AI DC integration, including load forecasting uncertainty, stability concerns, resource adequacy, and forced oscillation implications. Building on these challenges, I will discuss the spectrum of emerging research opportunities that span power systems engineering, data center architecture, control and optimization, and cross‑sectoral coordination. These opportunities highlight the need for new paradigms in flexible demand management, demand‑response mechanisms tailored to AI workloads, improved modeling of large computing clusters, and enhanced grid‑interactive data center designs. Finally, the talk will conclude with a deep dive into one specific subproblem within this broader landscape, illustrating current methodologies, open questions, and future research directions.