Battery storage is often pitched as the answer to the AI power problem. That is too broad. Batteries are not a substitute for a dependable energy supply, and a four-hour system does not turn a 100-megawatt campus into an island for a week.

The more useful question is what the battery is being asked to do. At an AI site, storage can bridge short outages, smooth ramp rates, reduce demand charges, provide reserve capacity, shift charging and discharging into lower-cost hours, and help a campus participate in grid programs. Its value comes from stacking those services without pretending that every function is available at the same time.

The basic sizing math

A battery has two headline ratings: power in megawatts and energy in megawatt-hours. A 50 MW / 200 MWh system can discharge at 50 MW for four hours under its stated conditions. It cannot provide 50 MW indefinitely.

An AI campus drawing 100 MW would use 200 MWh in two hours if the entire load were placed on the battery. In practice, the battery may serve only critical loads, and the data center may shed nonessential systems. The design therefore starts with a load hierarchy: cooling, networking, storage, control systems, safety systems, and selected compute loads.

The battery also needs reserve. Operators rarely plan to run to zero state of charge. They keep headroom for a second event, account for temperature, and limit charge and discharge rates to protect asset life. The usable figure is lower than the nameplate number.

The U.S. battery market is scaling quickly. EIA reported in August 2026 that operational utility-scale battery capacity reached 43.6 gigawatts by the end of 2025 and nearly 52 gigawatts after additions in the first half of 2026. EIA said developers reported another 54 gigawatts planned over the following two and a half years. That is a sign of a growing supply chain, not a guarantee that every proposed project will be built.

What batteries can solve at a data center

The first use is ride-through. A battery can cover the seconds or minutes needed for controls to respond and generators to start. That can reduce reliance on an uninterruptible power supply for longer events, or extend the time available to transfer loads.

The second use is peak management. If the utility charges for the highest monthly demand, a battery can discharge during a short peak and reduce the bill. The economics depend on tariff rules, the battery’s round-trip efficiency, degradation, and how often a peak can be avoided.

The third use is interconnection support. A campus may have a 100 MW utility service but a load that ramps quickly from 40 MW to 100 MW. Storage can slow the ramp or supply a portion of the difference while the utility studies a larger connection. It does not create new net energy, but it can help a site use its connection more intelligently.

The fourth use is market participation. In some regions, batteries can provide frequency regulation, reserve, or capacity. A data center owner should be careful: grid services may conflict with the site’s own reliability requirement. The contract needs a priority rule that protects the campus first.

Cost is not just the battery container

EIA reported a capacity-weighted average installed cost of $1,469 per kilowatt for battery systems installed in 2024. That is a broad utility-scale average. An AI-site project can cost more or less depending on duration, fire protection, interconnection, controls, site work, and whether it is located inside or outside the data-center campus.

The project also has replacement and augmentation costs. EIA’s capital-cost assumptions for a 150 MW / 600 MWh lithium-ion system include annual degradation of about 1.5% of initial battery capacity, with augmentation through added batteries. Real projects vary, but the principle is universal: the battery is a degrading asset, not a static generator.

Insurance, fire protection, monitoring, and emergency response are material. DOE’s Cybersecurity, Energy Security, and Emergency Response office warned in January 2025 that battery systems face supply-chain and security risks involving foreign-manufactured components, inverters, and transformers. Procurement should include software support, spare parts, remote access controls, and a plan for end-of-life modules.

Fire safety and siting change the design

Large lithium-ion systems are engineered around thermal runaway risk. The owner needs appropriate spacing, detection, suppression or mitigation systems, ventilation, drainage, emergency access, and a response plan coordinated with local authorities.

The battery may be placed outdoors in containers, in a dedicated building, or in a separated campus yard. Each choice affects noise, cable length, fire separation, weather exposure, and maintenance. The shortest electrical path is not always the best safety path.

A data center should also distinguish battery storage from UPS batteries. A UPS is designed to bridge and condition power for critical electronics. A utility-scale BESS is designed for larger energy movement and may be operated under a different control regime. They can work together, but they should not be treated as interchangeable boxes.

The right investment test

Start with the outage problem, tariff, and interconnection constraint. Then model the battery’s duty cycle, reserve, degradation, replacement, and insurance. Run a case with no energy-market revenue. If the project only works when every possible revenue stream is collected at maximum value, it is fragile.

Battery storage is most valuable when it gives an AI campus flexibility: time to start a generator, time to reduce noncritical load, time to recover from a grid disturbance, or the ability to shift consumption into a more favorable hour. It is not a replacement for firm generation, transmission, or a well-designed utility connection.

Small modular reactors occupy an awkward space in the data-center conversation. They are far enough along to attract customers, utilities, investors, and regulators, but early enough that many public timelines still describe milestones rather than delivered electricity.

That distinction is critical for anyone underwriting an AI campus. A reactor announcement can be real and still be irrelevant to a site’s 2027 power plan. The correct question is not whether a developer has a design. It is whether the project has a regulator-approved path, a site, a supply chain, a construction plan, a buyer, and a schedule with enough contingency to survive ordinary delays.

What “small modular” actually changes

An SMR is generally a reactor designed for smaller output and factory-style repetition compared with a large conventional nuclear plant. The label does not remove the need for a site, safety analysis, security, fuel, cooling, emergency planning, construction quality, and operating oversight.

The benefit is supposed to come from repeatability. A smaller unit can fit more sites, be manufactured in modules, and add capacity in steps. But repeatability only lowers cost after a design has been built, inspected, licensed, and produced more than once. The first unit carries the risk that later units are expected to avoid.

The Nuclear Regulatory Commission says advanced and microreactors can be licensed today under existing Part 50 and Part 52 frameworks. In 2026, the NRC also moved on additional frameworks: the Part 53 rule was published in the Federal Register on March 30, 2026 and became effective April 29, while a proposed Part 57 rule for microreactor security was published in draft form on May 1.

Those are meaningful regulatory developments, but they are not construction permits. They make the pathway clearer; they do not make a reactor site operational.

A schedule has at least five clocks

The first clock is design readiness. A developer must have a design detailed enough for safety analysis and licensing. The second is regulatory review. The third is site and environmental review. The fourth is procurement and manufacturing. The fifth is construction, commissioning, fuel loading, testing, and commercial operation.

A project can be advanced on one clock and early on another. A design might have a customer agreement while its fuel form is not yet produced at commercial scale. A company might have a site while its construction permit is still pending. A utility might announce a target decade while no long-lead equipment has been ordered.

The NRC highlighted licensing efficiencies in 2025, including a 17-month review for the TVA Clinch River SMR construction permit and an 18-month review milestone for the Dow Project Long Mott advanced reactor design. Those examples show that reviews can be accelerated, but they also show the scale of work that must be completed before a project is ready to build.

What data-center buyers should demand

A data-center buyer should request the exact project milestone, not a marketing label. Is it a memorandum of understanding, an offtake agreement, a site option, a docketed application, a construction permit, a combined license, or an operating license?

The buyer should also ask what the agreement actually obligates. A conditional offtake agreement can reserve future capacity without guaranteeing delivery. A power purchase agreement may depend on licensing, financing, interconnection, or a final investment decision. The buyer must know which conditions are precedent to performance and who bears delay risk.

Fuel is another overlooked issue. Advanced reactors may require fuels with specialized enrichment or fabrication capacity. A design can be technically ready and still wait for a commercial fuel chain. The project model should identify the fuel supplier, the qualification status, the delivery schedule, and the backup plan.

Why 2030 headlines can mislead a 2027 site

The IEA reported in 2025 that data centers are being built on a two-to-three-year technology cycle while the broader energy system requires longer planning, permitting, and construction periods. That mismatch explains why SMRs are discussed as an eventual answer even when a campus needs power now.

A project with a 2030 commercial target might be useful in a long-term energy strategy. It cannot support a 2027 lease unless the campus has firm interim power. The interim plan may include grid service, gas generation, batteries, or a different site.

The prudent approach is to value the reactor as an option until the project crosses objective milestones. Assign no capacity credit for an announcement. Assign partial credit for a signed, funded, licensed project with a site and procurement plan. Assign firm capacity only when there is a tested operating unit or a contract with enforceable replacement power.

The headlines still matter

Skepticism should not become dismissal. The NRC’s 2025 and 2026 actions reduce uncertainty around licensing, and the industry is attracting serious industrial customers. The IEA reported that conditional offtake agreements between data-center operators and SMR projects grew from 25 gigawatts at the end of 2024 to 45 gigawatts in its later 2026 update.

That growth signals demand for firm power. It does not prove that all 45 gigawatts will be built. The useful lesson is to watch conversion rates: how many conditional agreements become site permits, how many licensed projects reach construction, and how many units reach operation.

SMRs could become an important part of the AI power mix. For now, their value is clearest as a future option that should be paired with a separately financed near-term power plan. Buyers who understand the difference between a headline date and an energized megawatt will make better site decisions.

The data-center power story is often told as a choice between the grid and a new nuclear plant. In practice, the nearer-term choice is frequently more ordinary: bring a gas turbine onto the campus while the utility connection catches up.

That option is gaining attention because a large AI campus can be ready to fill racks before a transmission upgrade is complete. The turbine is not a magic source of cheap electricity. It is a project-management tool that converts a fuel connection and a permitted generator into usable megawatts. The trade is that the operator takes on fuel-price risk, local air permitting, equipment maintenance, and a more complicated relationship with the grid.

What a campus turbine actually does

A behind-the-meter turbine generates electricity on the customer side of the utility meter. The data center can run it as its primary supply, use it during constrained hours, or combine it with grid service and batteries. The arrangement can also be structured as a microgrid, with controls that isolate the campus during a utility outage.

A simple-cycle combustion turbine is attractive when speed and flexibility matter. It can start relatively quickly and is useful as a peaking resource, but its fuel efficiency is weaker than a combined-cycle plant. A combined-cycle plant adds a heat-recovery steam generator and steam turbine. That produces more electricity from the same gas, but it adds cost, equipment, water or air-cooling decisions, and construction time.

The distinction matters because a data center has a high and unusually steady load. A peaker built for a few hundred hours per year may be a poor fit if it must run around the clock. Conversely, a fast-start turbine can be valuable as a bridge while the permanent supply is built.

The Energy Information Administration reported in June 2025 that developers planned 18.7 gigawatts of new U.S. combined-cycle capacity through 2028, including 4.3 gigawatts already under construction. EIA also noted that more than half of the 3.3 gigawatts planned for 2026 was already under construction, while much of the 2027 pipeline was not. Those figures describe utility-scale projects, not campus generators, but they show the difference between a named project and equipment that is moving toward operation.

The first constraint is often the gas system

A turbine needs a firm fuel supply, not just a nearby pipeline on a map. The project owner must confirm pressure, delivery capacity, interconnection terms, curtailment rules, and whether the pipeline can support the expected hourly profile during cold weather.

This is especially important in New England and other regions where gas demand rises for heating. The power plant may be able to run in normal conditions and still face fuel constraints on the coldest days. A data center that promises nonstop service cannot treat gas as an assumed input. It needs a supply contract, backup fuel policy where permitted, and a plan for what happens if the pipeline operator curtails deliveries.

Fuel cost is another variable. EIA reported that Henry Hub gas averaged $3.52 per million British thermal units in 2025, 56% above 2024. Northeast hub prices averaged roughly twice 2024 levels. A turbine with a low capital cost can therefore become expensive to run when local gas prices spike.

Efficiency is a financial variable, not a footnote

The heat rate measures how much fuel a generator needs to produce one kilowatt-hour. EIA has reported that newer combined-cycle plants generally have heat rates below 7,000 Btu per kilowatt-hour, while simple-cycle and steam units are typically above 10,000 Btu per kilowatt-hour.

Consider a simplified example. A 100-megawatt turbine running at a 7,000 Btu per kilowatt-hour heat rate consumes about 700 million Btu per hour at full output. At $4 per million Btu, the fuel cost is about $2,800 per hour, before operations, maintenance, emissions costs, and capital recovery. At a 10,000 Btu heat rate, the same output uses about 1,000 million Btu per hour, raising the fuel bill to $4,000 per hour.

These are illustrative calculations, not a forecast. They show why the nameplate megawatt number is not enough. The financial model needs hourly dispatch, expected load factor, heat rate at partial load, start costs, maintenance intervals, fuel transportation, emissions costs, and the cost of purchased grid power.

Permits and community acceptance can set the schedule

A campus turbine is still a power plant. It may need air permits for nitrogen oxides, carbon monoxide, particulate matter, greenhouse gases, and startup or emergency operation. Noise, visible exhaust, water use, and emergency fuel storage can all affect local review.

The fast path is not always the cheapest path. A permit that limits operating hours may undermine the reason for installing the turbine. A requirement for selective catalytic reduction can add cost and water or reagent logistics. Diesel backup generators may exist for reliability but are not automatically an acceptable substitute for a continuous primary supply.

Operators should publish a clear operating envelope: expected annual hours, normal dispatch, emergency dispatch, fuel type, emissions controls, and how the project will respond to grid emergencies. That turns a technical project into a reviewable public commitment.

The prudent model is usually hybrid

The strongest case for campus turbines is often not grid replacement. It is staged power.

Phase one uses temporary or bridge generation to energize an initial building. Phase two adds utility service and perhaps battery storage. Phase three moves the site toward a lower-carbon mix through renewable contracts, firm clean power, or a future nuclear or geothermal resource.

This structure also protects against stranded equipment. If the grid connection arrives early, the turbine can become a resilience and peak-management asset instead of an oversized primary plant. If the connection is delayed, it can keep the site moving.

The key diligence question is simple: what problem is the turbine solving? If it solves a documented interconnection delay and the fuel, permits, and operating costs are underwritten, it can be a practical bridge. If it is being used to make an uncertain site look powered on a slide deck, the risk is much higher.