From Coal to Carbon Neutral: The AI Playbook

A small campus in North Carolina is doing what many large organizations still treat as “too hard”: running modern buildings, keeping people comfortable, and cutting emissions—at the same time. Catawba College, with roughly 1,200 students, earned certified carbon neutral status and is now modernizing aging infrastructure with a strategy that’s both practical and replicable.

If you work in energy and utilities, the interesting part isn’t the campus marketing headline. It’s the operating model underneath it: district energy + geoexchange + solar + high-performance buildings, all sequenced around deferred maintenance and long-term costs. My take: Catawba’s approach is exactly the kind of real-world “messy system” where AI for energy management stops being theory and starts paying rent.

What follows is a case-study-based playbook for utilities, ESCOs, and large energy users (higher ed, healthcare, local government, industry) who want to scale decarbonization without betting the farm.

Why Catawba’s model works (and why most don’t)

Answer first: Catawba succeeds because it treats decarbonization as an infrastructure modernization program, not a collection of green projects.

Most organizations get stuck in one of two traps:

1. “Add renewables and call it done.” You end up with solar on a campus that still wastes energy through leaky envelopes, oversized equipment, and manual controls.
2. “Fix deferred maintenance first.” You replace equipment like-for-like and lock in fossil dependence for another 20–30 years.

Catawba threads the needle by aligning decarbonization with the stuff boards already fund:

• Replacing aging HVAC and central plant assets
• Improving occupant comfort (humidity, ventilation, hot/cold complaints)
• Reducing exposure to volatile energy prices
• Creating a campus “living lab” that supports recruitment and retention

That alignment matters because it changes the internal question from “Can we afford sustainability?” to “Can we afford not to modernize the right way?”

The backbone: district energy + geoexchange as a platform

Answer first: The fastest path off fossil heat on many campuses is district energy modernization anchored by geoexchange (ground-source heat) and electrified distribution.

Catawba’s District Energy and Modernization (DEM) project centers on a closed-loop geoexchange buildout. In 2024, the college drilled 39 wells as part of a four-phase system designed to ultimately serve 26 campus buildings. The project is projected to cut at least 265 MMBtu of gas and 145,724 kWh of purchased electricity per year, avoiding about 51 metric tons of carbon emissions.

Two practical lessons jump out for energy and utility teams:

1) Treat thermal as “the other grid”

Electric utilities think naturally in networks—feeders, substations, contingencies. Campuses and districts often don’t apply that systems thinking to heating and cooling.

Geoexchange-based district energy turns heating/cooling into a networked asset with properties utilities care about:

• Predictable performance (ground temperatures are stable)
• Centralized maintenance (especially when wells are placed strategically, like under a soccer field)
• Upgradeable endpoints (heat pumps and building-side equipment can be swapped over time)

2) Legacy assets can become decarbonization accelerators

Catawba benefits from earlier geoexchange investments: some older buildings are already connected, making new hookups cheaper.

That’s a broader point: an imperfect legacy system can still be a head start if you map it, instrument it, and modernize around it.

Where AI fits: the “controls gap” that makes or breaks ROI

Answer first: AI creates value when you have electrified, distributed assets (heat pumps, solar, storage, district loops) and need to run them as one coordinated system.

Catawba’s physical upgrades are the foundation. But the big, scalable opportunity—especially for utilities and energy service providers—is operational intelligence:

AI use case #1: Campus energy optimization (multi-objective control)

Once you have geoexchange loops and multiple buildings, the hard part is deciding how to operate:

• When do you pre-heat or pre-cool?
• How do you avoid demand spikes during morning warm-up?
• How do you prioritize comfort in sensitive spaces (gyms, pools, labs) without wasting energy?

AI-enabled optimization can balance competing targets:

• Cost (tariffs, demand charges, time-of-use)
• Carbon (marginal emissions signals)
• Comfort (temperature, humidity, IAQ constraints)
• Equipment health (runtime balancing, short-cycling avoidance)

A plain-language definition you can use internally:

AI energy optimization is automated decision-making that minimizes cost and emissions while meeting comfort and equipment constraints.

AI use case #2: Demand forecasting for electrified heat

Electrifying heat shifts load shapes. Heating demand can be peaky, weather-driven, and dependent on building envelope quality.

AI forecasting helps in three places:

• Day-ahead scheduling: anticipate cold snaps and stage equipment
• Real-time dispatch: prevent peak events by smoothing starts
• Procurement and hedging: reduce exposure to price volatility by predicting load more accurately

For utilities, this is where “campus micro-grid behavior” becomes grid-relevant behavior, especially as electrification scales across regions.

AI use case #3: Predictive maintenance for heat pumps, pumps, and wells

Catawba’s story includes a key operational reality: modern systems are efficient, but they’re also more sensor-rich and control-dependent.

Predictive maintenance models can detect:

• Pump degradation (vibration, power draw anomalies)
• Heat pump performance drift (COP decline, refrigerant issues)
• Fouling or flow problems (ΔT patterns that don’t match expected behavior)

This isn’t academic. Predictive maintenance is what keeps high-performance systems from sliding into “runs, but inefficient” mode—the silent killer of projected savings.

High-performance buildings: why envelopes beat heroics

Answer first: You can’t control your way out of a bad envelope; efficiency is still the cheapest decarbonization fuel.

Catawba’s modernization highlights a common facilities problem: older buildings with inadequate insulation, vapor barriers, and inconsistent HVAC upgrades. One example is the 90,000-square-foot Abernethy Physical Education complex, which includes a pool and has had comfort challenges in humid North Carolina conditions.

One quote from the project team nails it: if you fix HVAC without fixing the envelope, you’re “heating and cooling the outside.”

Catawba’s new construction leans into this reality. In 2025, the college began building a 130-bed residence hall designed for Passive House US certification. Compared to typical North Carolina construction, the college expects:

• 55% lower energy use
• 44% lower heating loads
• 75% lower cooling loads

Here’s the AI connection that many teams miss: better envelopes make AI control better. When buildings respond predictably, optimization algorithms perform more reliably and occupant complaints drop.

Symbolism that actually matters: repurposing the coal plant

Answer first: Turning a coal-fired power plant into a net-zero (or better) building isn’t just a PR win—it’s an asset strategy.

Catawba’s Smokestack building, once a coal plant (1950s–1990s), is being converted into a 10,000-square-foot student life hub that will connect to district energy, use photovoltaics, and target net-zero or net-positive energy performance. It’s slated for LEED Platinum and alignment with the Living Building Challenge Core Imperative, which broadens the lens to water, materials, ecology, and education.

From an energy-and-utilities perspective, adaptive reuse also reduces the need for new embodied carbon-heavy construction—often the forgotten half of “net zero.”

A practical playbook utilities can reuse with customers

Answer first: The repeatable pattern is Assess → Modernize thermal → Tighten envelopes → Electrify → Optimize with AI → Verify savings.

If you’re building programs for large energy users (campuses, cities, hospitals), Catawba’s sequence translates into an implementation checklist.

Step 1: Build a “decarbonization + deferred maintenance” portfolio

Stop proposing single projects. Propose a funded roadmap where projects reinforce each other.

• Bundle end-of-life equipment replacement with electrification
• Tie occupant comfort KPIs to energy KPIs
• Use NPV, not simple payback, to compare options

Catawba’s preliminary NPV analysis projects about $10 million in savings from 2025 to 2054 versus business as usual when factoring energy, operations, and maintenance.

Step 2: Instrument first, then automate

Before advanced optimization, you need trustworthy data:

• Submetering by building and major loads
• Supply/return temperatures and flows for district loops
• Heat pump performance data (runtime, setpoints, alarms)
• Occupancy proxies (schedules, badge counts, CO₂ sensors where appropriate)

Then apply AI where it belongs: in dispatch, forecasting, fault detection, and measurement & verification.

Step 3: Put carbon on the dashboard—not in a PDF

If the goal is carbon neutrality, operations teams need carbon signals in daily workflows:

• Operational emissions by hour/day
• Avoided emissions from solar and load shifting
• Marginal emissions-informed control strategies (when available)

A simple stance: if carbon isn’t visible to operators, it won’t be managed.

Step 4: Use contracts that reward outcomes

Utilities and ESCOs can accelerate adoption with performance structures:

• Shared savings models tied to verified kWh/therms and peak demand
• Comfort guarantees (temperature/humidity bands)
• Uptime SLAs for critical facilities

AI helps here too: better verification, faster fault detection, fewer disputes.

People also ask: what makes carbon neutrality achievable for small institutions?

Answer first: Carbon neutrality becomes achievable when institutions combine deep efficiency, electrified thermal systems, and renewable generation—and then run it well.

Catawba’s example shows that scale isn’t the deciding factor. The deciding factors are:

• Leadership willing to fund modernization strategically
• Projects designed as a system (not one-offs)
• Financing framed around long-term stewardship and risk reduction
• A culture of measurement and learning

What to do next if you’re a utility, ESCO, or energy leader

Catawba College proves a point I wish more decarbonization plans would internalize: the work isn’t choosing between comfort, cost, and carbon. The work is designing for all three. And once the physical system is modernized, AI becomes the force multiplier—turning a collection of assets into an optimized operation.

If you’re building your 2026 program pipeline right now, look for customers facing large deferred maintenance waves (higher ed is full of them) and propose an integrated roadmap:

• District thermal modernization anchored by geoexchange and heat pumps
• High-performance envelope upgrades where they matter most
• Solar and storage sized to reduce peaks, not just annual kWh
• AI-driven forecasting, optimization, and predictive maintenance to keep savings real

The next question is the one every energy leader should be asking heading into 2026 budgeting cycles: Which parts of your system are you still running “by feel” that should be run by data—and which parts of your data aren’t good enough to trust yet?