AI-Ready Decarbonization: Lessons From Catawba

A 1,200-student college in North Carolina doesn’t sound like the place you’d look for infrastructure strategy. Yet Catawba College became one of only 13 institutions in the US certified carbon neutral—and it did it while tackling the unglamorous stuff most organizations avoid: deferred maintenance, aging HVAC, and outdated buildings.

That combination is exactly why this story belongs in an AI in Energy & Utilities series. Utilities and energy leaders know the hard truth: decarbonization doesn’t fail because people dislike clean energy. It fails because assets are old, capital is constrained, and operational complexity is real. Catawba’s approach—pairing renewables with district energy modernization and high-performance buildings—maps cleanly to what AI enables across the energy sector: better planning, better operations, and fewer expensive surprises.

Catawba didn’t “buy” carbon neutrality with offsets and good intentions. It built the operational backbone for it: solar, geoexchange, and a district energy strategy designed to reduce fossil fuel exposure and stabilize long-term costs. The opportunity for energy and utility teams is to treat this case study as a template—and then add AI where it naturally belongs.

Catawba’s blueprint: modernize infrastructure and cut emissions

Catawba’s core move is simple: align decarbonization with unavoidable reinvestment. The college is upgrading buildings and systems that were going to need work anyway, but it’s doing it in a coordinated way that reduces energy use and emissions instead of locking in another 20 years of patchwork.

The campus strategy includes:

• On-site solar PV: ~837 kW installed in 2015 plus an additional 55 kW added in 2024, with further expansion planned.
• Geoexchange (ground-source) heating and cooling: first installed in the 1990s, now expanding into a campus-wide district system.
• High-performance buildings: a coal-era facility repurposed as a net-zero (or better) student hub, plus new construction designed to Passive House standards.

What’s notable for energy leaders is that none of these are exotic technologies. The value comes from sequencing and integration.

The underappreciated win: tying deferred maintenance to decarbonization

Most organizations treat deferred maintenance as a separate budget category—something you “catch up on” when you can. Catawba treats it as a decarbonization accelerator.

Aging systems (especially HVAC) create a trap:

• Replace a chiller without improving the building envelope, and you’ll oversize equipment and waste energy.
• Add insulation without modern controls, and comfort complaints still pile up.
• Add solar without load optimization, and you risk curtailment or poor economics.

Catawba’s project teams explicitly call out this systems logic. As one project leader put it: if you fix HVAC without fixing the envelope, you’re “heating and cooling the outside.” That’s not just a building-science quip—it’s a governance lesson. Decarbonization needs portfolio-level coordination, not building-by-building heroics.

District energy modernization + AI: where the biggest operational gains hide

Catawba’s District Energy and Modernization (DEM) project is a practical example of how to reduce gas dependence without betting the farm on a single building.

In 2024, Catawba drilled 39 wells for a closed-loop geoexchange system—part of a four-phase plan to ultimately serve 26 campus buildings. Once complete, the system is expected to cut at least 265 MMBtu of gas and 145,724 kWh of purchased electricity per year, reducing carbon emissions by 51 metric tons.

Here’s the AI connection: geoexchange and district energy systems are data-rich and optimization-heavy. That’s good news, because AI excels when operations involve many variables and constraints.

How AI strengthens district energy performance

If you’re running district energy (on a campus, in a city, or across a utility service territory), you’re constantly balancing comfort, efficiency, peak demand, and equipment health. AI can help in three high-ROI ways:

1. Predictive load forecasting (15-minute to multi-day horizons)
   AI models can forecast heating/cooling demand using weather, occupancy schedules, and historical patterns. That enables smarter setpoints, lower peaks, and less cycling.

2. Optimal dispatch across assets
   When you have geoexchange, heat pumps, backup boilers, thermal storage, and solar, the “best” combination changes hour by hour. AI-driven optimization can minimize cost or emissions (or both) under real constraints.

3. Continuous commissioning
   The fastest energy savings often come from finding what drifted: stuck dampers, failing sensors, mis-tuned control loops, or simultaneous heating and cooling. Machine learning anomaly detection can flag issues early—before humans notice comfort complaints.

The stance I’ll take: district energy without data discipline becomes a comfort project, not a decarbonization project. Catawba’s integrated approach sets it up for the next step—instrumentation, analytics, and AI-assisted operations.

From coal plant to net-zero hub: why adaptive reuse matters to utilities

Catawba’s “Smokestack” project takes a coal-fired power plant (operational from the 1950s to 1990s) and turns it into a 10,000-square-foot student life hub and living laboratory, designed to connect to the district energy system and use photovoltaics—reaching net-zero (or better).

For utilities and energy executives, adaptive reuse isn’t just an architectural story. It’s a mindset shift:

• The energy transition is full of legacy assets.
• Replacing everything is expensive and slow.
• Repurposing infrastructure can deliver faster emissions cuts with less political friction.

In the same way a college can transform an old plant into a high-performance facility, utilities can:

• Repurpose older substations with modern automation and sensors.
• Upgrade feeder lines and protection with AI-assisted fault detection.
• Extend asset life safely using predictive maintenance rather than calendar-based replacement.

Catawba is proving something many energy organizations need to internalize: legacy doesn’t have to mean liability.

High-performance buildings + smart controls: the demand-side “power plant”

Catawba’s new residence hall (groundbreaking in April 2025) is designed for Passive House US certification and will tie into the district system. Compared to typical North Carolina construction, the college estimates it will reduce:

• Energy use by 55%
• Heating loads by 44%
• Cooling loads by 75%

Those numbers matter because they translate directly into grid realities. Lower peaks and lower cooling loads reduce:

• summer capacity strain,
• transformer overload risk,
• and the need for peaker generation.

Where AI fits on the demand side (and why “BMS data” isn’t enough)

Many organizations already have building management systems (BMS). The problem is that BMS platforms often collect data without turning it into decisions.

AI turns building performance into a controllable asset by enabling:

• Occupancy-aware ventilation and temperature control (especially useful during holidays and exam weeks)
• Dynamic demand response tied to utility price signals or campus peak thresholds
• Fault detection and diagnostics that surface actionable work orders (not just alarms)
• Envelope and HVAC co-optimization, preventing situations where efficient equipment is wasted by leaky walls or poor humidity control

If you’re in a utility role, the lesson is broader: efficient buildings are grid infrastructure. Treating them that way—through incentives, performance contracts, and AI-enabled measurement and verification—creates a reliable “virtual capacity” resource.

Financing and risk: why the “small institution” angle matters

Catawba’s leadership framed the DEM investment as more than ROI: recruitment, retention, carbon goals, and deferred maintenance. Still, the economics are compelling. A preliminary net present value analysis suggests the DEM approach could save about $10 million from 2025 to 2054 compared to business-as-usual, considering energy, operations, maintenance, and related costs.

For energy and utilities teams trying to generate leads and internal buy-in, this is the persuasive framing:

• Decarbonization reduces exposure to volatile fuel prices.
• Modernization reduces unplanned outages and emergency repairs.
• Comfort and reliability reduce “soft costs”—complaints, churn, staffing strain, and reputational hits.

A practical “AI + infrastructure” roadmap (steal this)

Catawba’s case suggests a roadmap smaller organizations can actually execute—especially if they partner with utilities, ESCOs, or energy service providers.

1. Instrument the critical assets first
   Start with meters, submetering, and sensor validation for major loads (chilled water, heat pumps, air handlers, PV output).

2. Build a clean data layer
   Normalize naming conventions, timestamps, and units. Bad tags ruin analytics.

3. Deploy two AI use cases before you deploy ten
   Pick high-confidence wins:

   • predictive maintenance for HVAC and pumps
   • anomaly detection for energy drift and comfort faults

4. Integrate optimization only after trust is earned
   Once operators see reliable insights, move to automated setpoint optimization and dispatch.

5. Measure outcomes in operational terms
   Track peak kW reductions, avoided maintenance events, comfort KPIs, and emissions—not just energy use.

This matters because too many “AI in energy” initiatives start with a platform purchase and end with dashboards nobody checks.

What utilities and energy leaders should take from Catawba

Catawba College’s story is a clean-energy transition case study with unusually high signal for utilities: it connects renewable integration, energy efficiency upgrades, and infrastructure modernization into one plan. That’s the same integration challenge utilities face, just at a different scale.

The most transferable lesson is blunt: you don’t decarbonize by adding clean tech onto messy operations. You decarbonize by modernizing the system. AI makes that modernization faster and more reliable—through forecasting, optimization, and predictive maintenance.

If you’re planning a district energy program, a campus microgrid, a utility demand-side portfolio, or a multi-site decarbonization effort, the next question isn’t “Should we use AI?” It’s: Which operational decisions are we still making blind, and what would it cost us to keep guessing for another five years?