Data centers are driving AI and also emissions. Here’s how smart permitting, diesel phase-outs, and clean backup power can make them truly green infrastructure.
Most companies chasing AI capacity are also quietly signing up for 300%+ growth in energy use, water withdrawals, and local air pollution from data centers.
That’s not a side story. It’s the story.
Here’s the thing about green technology: it only counts as green if the infrastructure behind it is clean too. California’s latest data center analysis makes that painfully clear — and it also points to a practical path forward that other regions can copy.
This post breaks down what California is learning about data center pollution, why diesel backup is becoming a public health problem, and how smart permitting and clean backup power can turn data centers into real assets for climate and communities.
The hidden environmental bill of AI data centers
The key point: data centers are growing far faster than most cities or utilities are prepared for, and the environmental tab is exploding alongside them.
Researchers from the University of California, Riverside and policy group Next 10 looked at data centers in California — already one of the cleanest grids in the U.S. — and the numbers are blunt:
- Electricity use from California data centers jumped 95% between 2019 and 2023.
- By 2028, it could be up 356% from 2019, hitting 25.3 TWh per year — about as much power as 2.4 million households.
- Carbon emissions from data centers nearly doubled in four years, from 1.24 million tons (2019) to 2.38 million tons (2023), and could reach 5.56 million tons by 2028.
- Cooling and power generation could consume up to 116 billion liters of water annually, in a state already dealing with drought.
This matters because California is one of the best-case scenarios: roughly two‑thirds of its electricity already comes from clean sources. If data center pollution is surging there, the picture is worse in regions running heavily on coal and gas.
Green technology advocates like to talk about AI accelerating clean energy, smarter grids, and efficient buildings. All true. But if those very AI workloads are hosted in facilities backed by diesel, guzzling water, and driving asthma cases, we’re just moving the damage around.
The reality? The problem isn’t data centers themselves. It’s where and how we build them — and the policies that either reward or ignore sustainable design.
Diesel backup: the dirtiest part of ‘always on’ infrastructure
If you care about public health, diesel backup generators are the red flag.
Around 90% of California’s data centers rely on diesel generators for backup power. In theory, those engines sit idle and only run during outages. In practice, they:
- Are tested regularly to meet air-quality rules and reliability standards
- Are used to manage peak loads when the grid is strained
That means ongoing emissions, not rare exceptions.
Those generators release fine particulate matter (PM2.5) and nitrogen oxides (NOx) — exactly the pollutants linked to asthma, cardiovascular disease, and premature death. The UC Riverside report tracks what that means in economic terms:
- Health-related costs tied to data center emissions in California tripled from $45 billion in 2019 to $155 billion in 2023.
- By 2028, those costs could hit $267 billion.
- Even under a relatively low-growth scenario, emissions would be responsible for roughly 3,900 asthma symptom cases and 490 lost workdays per year by 2028.
Those aren’t abstract “social costs.” They’re real hospital visits, missed paychecks, and pressure on already stretched healthcare systems. And like most pollution, the burden doesn’t fall evenly. Industrial zones hosting big clusters of backup generators are often near lower‑income or historically marginalized communities.
If you’re a city planner, utility, or a company serious about ESG, staying on diesel is increasingly hard to justify.
Smart permitting: using approvals to steer greener data centers
Smart permitting is the most powerful, underused tool cities have to clean up data center growth.
The idea is simple: speed up and streamline approvals for projects that are energy-efficient, water-smart, and located in the right places — and make it harder or slower for those that aren’t.
From the California analysis, a “smart permitting” framework for data centers looks something like this:
1. Reward the right locations
Not every grid node or watershed can handle a hyperscale facility. Permitting should prioritize sites that:
- Sit near abundant clean power (strong renewables, transmission capacity, low congestion)
- Have reliable, sustainable water supplies and/or non‑potable sources for cooling
- Aren’t in air basins already failing to meet health-based standards
Fast-track pathways, predictable timelines, and coordinated reviews can turn those preferred zones into green data center districts, reducing uncertainty for developers and protecting communities.
2. Bake efficiency into the approval process
Permits shouldn’t just ask, “Is this allowed here?” They should ask, “Is this facility designed to be among the most efficient in its class?”
You can hard‑wire that into rules by tying approvals to:
- Aggressive power usage effectiveness (PUE) targets
- Use of free cooling, liquid cooling, or other efficient thermal strategies
- Heat recovery plans (e.g., exporting waste heat to district heating where practical)
- Use of non‑potable water, recycled water, or hybrid cooling modes to cut freshwater withdrawals
If a facility hits those performance thresholds, it gets faster, clearer permitting. If not, it faces more scrutiny, mitigation requirements, or simply doesn’t get approved.
3. Connect permitting to clean backup power
One of the strongest levers is to phase out diesel backup generators via the permit system:
- New large data centers: require non‑diesel backup (batteries, fuel cells using low‑carbon fuels, or other clean options) as a condition of approval
- Existing sites: set schedules where permit renewals are tied to diesel phase-down milestones
It’s unreasonable to just declare diesel “banned tomorrow.” But tying continued operation and expansion to a clear transition path is both fair and effective.
For regulators and cities, smart permitting is appealing for another reason: you don’t have to outlaw data centers or pick winners. You just align approvals with your climate, water, and air quality goals and let capital flow where it fits.
Clean backup power: practical alternatives to diesel
If diesel is the problem, what replaces it without sacrificing reliability? The good news for green technology is that the alternatives are no longer theoretical.
Battery energy storage systems (BESS)
Modern battery systems can already take over many roles diesel plays:
- Instant response to grid disturbances
- Ride‑through for short‑term outages
- Support for demand response and grid services
Paired with on‑site solar or long‑term renewable power contracts, BESS turns a data center from a passive load into a flexible grid resource. The main constraints are duration and cost, but for most urban and suburban sites — where outages are measured in minutes or hours, not days — batteries are enough.
Fuel cells and low‑carbon fuels
Fuel cells running on green hydrogen, biogas, or other low‑carbon fuels can provide longer-duration backup at much lower local emissions than diesel. They also offer:
- High efficiency
- Quiet operation
- Modular scalability
The catch: fuel supply chains and infrastructure are still maturing. That’s where policy and permitting can help by giving long-term visibility to developers willing to commit.
Hybrid backup architectures
The most realistic near‑term path in many regions will be hybrid systems:
- Batteries handle short events and grid services
- Cleaner thermal generators or fuel cells cover extended outages
- Diesel is retained only as a last‑resort layer, then phased out entirely as alternatives prove themselves
From a risk perspective, this layered approach makes sense. From a climate and health perspective, it sharply reduces hours of diesel operation, even before full phase-out.
If you’re responsible for data center design or ESG strategy, the direction of travel is clear: start building the business case for non‑diesel backup now, because regulations and community pushback are moving faster than many boards expect.
Data, not guesses: why standardized reporting is non‑negotiable
You can’t manage what you hide.
The UC Riverside report calls out a problem most regulators quietly complain about: there’s no consistent, public, standardized reporting framework for data center impacts across power, carbon, water, and health.
For a truly green technology ecosystem, that has to change.
A sensible reporting regime would require large data centers to disclose on a standardized basis:
- Annual electricity consumption and peak demand
- Carbon emissions, broken out by scope and time of use (since off‑peak may be cleaner)
- Water consumption by source (potable, non‑potable, recycled)
- Backup generator operation hours, fuel use, and local pollutant emissions
Once you have clean, comparable data, several things become possible:
- Regulators can target rules where they’d have the biggest impact.
- Cities can plan infrastructure around real loads instead of guesses.
- Companies can benchmark their facilities and compete on efficiency, not just capacity.
- Communities can actually see what’s happening in their backyard.
For AI‑driven green tech, this transparency is gold. Better data feeds better models — whether you’re forecasting grid stress, optimizing cooling, or mapping pollution hotspots.
Community science and cross‑sector collaboration
One of the more forward‑thinking recommendations from the California work is to institutionalize community science and cross‑sector research.
That means:
- Involving residents in monitoring local air quality and noise
- Making environmental and health findings available through plain‑language reports
- Running public workshops where researchers, regulators, and communities co‑design mitigation strategies
I’ve found that when communities see real numbers and are invited into the process early — instead of learning about a 200‑MW data center from a construction fence — the conversation changes. Companies get clearer social license to operate, and residents get more than vague promises.
For green technology as a whole, this is the model we need: not just cleaner hardware, but transparent, participatory governance around it.
What cities, utilities, and companies should do next
If you’re working in smart cities, energy, or sustainability, here’s a practical way to respond to this data center surge.
For cities and regions
- Create smart permitting frameworks that:
- Prioritize low‑impact locations
- Tie approvals to efficiency and non‑diesel backup
- Require transparent environmental reporting
- Coordinate across departments (planning, air quality, water, energy) so data centers go through one integrated, predictable process.
- Engage communities early, especially where multiple facilities may cluster.
For utilities and grid planners
- Treat data centers as flexible grid partners, not just giant loads.
- Design tariffs and programs that reward:
- Shiftable workloads
- On‑site storage
- Support for frequency and capacity services
- Use AI and advanced forecasting to map where new data center clusters align best with your clean energy build‑out.
For data center developers and AI leaders
- Commit to phasing out diesel backup on a defined timeline.
- Invest in battery storage, fuel cells, and on‑site renewables now, before policy forces rushed retrofits.
- Publish clear sustainability metrics for each site: power, carbon, water, and local air pollutants.
- Treat environmental performance not as an afterthought, but as part of your value proposition to customers who care about green technology.
The future of green tech depends on what powers it
Green technology isn’t just wind farms, solar panels, or smart thermostats. It’s the entire stack — including the data centers that make AI, cloud, and digital services possible.
California’s experience shows two truths at once:
- Data centers, left on autopilot, can blow through climate goals, strain water supplies, and raise healthcare costs.
- With smart permitting, non‑diesel backup, and transparent reporting, they can align with clean energy systems and even support grid reliability.
The question for every region racing to attract AI and cloud investment is simple: Are you just chasing megawatts and tax revenue, or are you designing a data center strategy that actually fits your climate and health goals?
The choices made over the next few years will define whether AI becomes a genuine pillar of green technology — or just another energy‑hungry industry that someone else has to clean up later.