The $282 Billion AI Problem: How Stevens Students are Mapping the Future of Data Centers
the staff of the Ridgewood blog
Hoboken NJ, As tech giants like Amazon, Google, Meta, and Microsoft pour an estimated $282 billion into the physical infrastructure of artificial intelligence, a massive problem is looming: where do we put it all?
Joseph Helsing (at left), lecturer in the Department of Electrical and Computer Engineering at Stevens, served as advisor to the data center siting team, which included (from left) Atif Qadir, Xirui Yu, Shivam Raj and Titir Talukder.
AI data centers aren’t your typical office buildings. They consume at least three times more energy than traditional facilities, straining power grids and sparking community pushback. Until now, many “siting” decisions were based on gut feeling or which town could rubber-stamp a permit the fastest.
Enter a team of graduate students from Stevens Institute of Technology. They’ve developed a machine-learning model to replace “intuition” with evidence, mapping the future of AI infrastructure across more than 85,000 U.S. Census tracts.
Moving Beyond “Gut Feeling” in Real Estate
The team, featuring Applied AI students Shivam Raj ’27, Titir Talukder ’27, Xirui Yu ’26, and real estate developer Atif Qadir, recognized that the current “land grab” for data centers is unsustainable.
“Today’s siting decisions will shape energy and technology systems for decades,” says Qadir. “But they’re often being made through intuition or by which region can approve permits the fastest.”
By consulting experts in construction, energy regulation, and urban planning, the Stevens team identified the key “performance-based factors” that make a site actually viable long-term.
How the Stevens Model Works
Using a sophisticated machine-learning approach, the team created a scoring system (0 to 100) for every Census tract in the United States. Unlike traditional models, their system learns patterns directly from data:
-
Proximity to Power: Analyzing high-voltage transmission facilities.
-
Connectivity: Tracking distance to major internet exchange points.
-
Environmental Risk: Factoring in climate and disaster data.
-
Policy Conditions: Evaluating local regulations and incentives.
The result? A data-driven map that shows exactly where the next “AI hubs” should be built to avoid grid collapse and maximize efficiency.
The Real-World Impact
Presented at the Stevens DuckAI Symposium in December 2025, the project is moving toward a future where real-time electricity pricing and grid expansion scenarios are baked into the model.
For student Titir Talukder, the goal is clear: “Understanding how and where data centers should be located helps support this expanding AI ecosystem while promoting sustainable, data-driven decision-making.“
As AI becomes indispensable in healthcare, finance, and everyday life, the work being done at Stevens ensures that the “brain” of AI has a sustainable place to live.
Take the Wall Street Walking Tour https://www.facebook.com/unofficialwallstreet #WallStreetTours,#FinancialDistrictExploration, #ExploreWallStreet, #FinancialHistoryTour, #StockMarketExperience, #FinancialDistrictDiscovery, #NYCFinanceTour,#WallStreetAdventure
