AI data centers are becoming one of the fastest-growing sources of electricity demand in the United States, and the race to power them is colliding with the country’s clean-energy ambitions. A typical data center once needed about 50 megawatts of power. AI facilities now routinely require 500 megawatts or more, roughly the load of a major city such as San Francisco. That surge has pushed utilities, developers, and technology companies into a familiar default: build more natural gas generation because it can be planned, permitted, and financed in predictable ways.

Firma Power, a startup founded by energy experts, argues that the gas default is not primarily a story about cost or reliability. The bigger obstacle is structural. Clean energy is abundant in the development pipeline, but data centers struggle to assemble enough of it from a scattered mix of wind, solar, and battery projects. Each project may have a different developer, timeline, interconnection status, and contract structure. Firma’s answer is software that treats those projects as a portfolio, matches them to a data center’s load, and negotiates a single power contract instead of a tangle of separate deals.

Key facts

• AI data centers can require 500 megawatts or more, compared with roughly 50 megawatts for a typical traditional data center.
• The U.S. is on track to add a record 86 gigawatts of new wind, solar, and battery capacity this year, but many projects still lack buyers or face grid and permitting delays.
• Firma Power’s FirmaOS platform identifies proposed clean-energy projects and matches them to a data center’s needs in minutes rather than months.
• The startup negotiates with multiple developers and gives the customer one contract, aiming to bring clean power online faster and at a competitive cost versus gas.
• Firma was founded by energy experts Jesse Jenkins, Scott Burger, and Brent Alderfer; it raised about $1.4 million in February 2026 and is working on roughly a dozen projects with hyperscalers and AI cloud companies.
• Firma estimates that three grid regions alone could add 200 gigawatts of clean energy by 2030, enough to support large-scale load growth if projects can be commercialized.

Why AI load growth changes the power conversation

For years, U.S. electricity demand was relatively flat. Efficiency gains, slower industrial growth, and modest population shifts kept many utilities from planning for large new generation additions. The rise of AI computing has disrupted that assumption. Data-center campuses are no longer modest warehouses of servers. They are industrial-scale power users, often built in clusters that can add hundreds of megawatts to a single grid in a short period. When a single facility can consume as much power as a city, the normal utility planning cycle starts to look too slow.

The problem is not simply that AI needs more electricity. It is that the electricity must arrive quickly, reliably, and in large enough blocks to satisfy companies that are racing to deploy models and cloud services. Natural gas plants have been the default answer because they can provide firm capacity and because the playbook for building them is well understood by utilities, regulators, and financiers. Clean-energy projects, by contrast, often arrive as individual wind farms, solar arrays, or battery installations, each with its own development risk. A data center that needs 500 megawatts cannot easily stitch together a dozen separate projects without taking on a dozen separate negotiations and timelines.

The clean-energy pipeline is large but fragmented

The good news is that clean power is not scarce in the abstract. The U.S. is poised to add a record 86 gigawatts of new wind, solar, and battery capacity this year. That is a substantial volume of new generation and storage, enough to serve millions of homes and a significant share of new commercial load. Yet a large pipeline does not automatically translate into power for AI data centers. Many projects sit in development for years before they secure buyers, finalize interconnection agreements, and move through local permitting. Some never reach construction because the economics do not line up, the grid operator cannot accommodate them, or the project cannot find a creditworthy customer.

This mismatch is at the heart of Firma’s pitch. A data center needs a reliable, long-term power supply. A clean-energy developer needs a buyer that can support financing. In theory, the two sides should be natural partners. In practice, they are separated by a fragmented market. Wind and solar produce variable output, so a buyer may need storage or a diversified portfolio to shape the supply. Projects are located in different grid regions, which raises transmission and deliverability questions. Contracts can vary in length, pricing structure, and risk allocation. The result is that gas often wins by default, not because it is always cheaper or cleaner, but because it is simpler to procure at scale.

How FirmaOS turns scattered projects into a contract

Firma Power’s platform, FirmaOS, is designed to reduce that complexity. The software looks at proposed clean-energy projects and calculates what combination of wind, solar, and storage could meet a data center’s capacity and energy needs. It then matches those projects to the facility’s load profile. According to the company, what might take months of manual analysis and negotiation can be done in minutes. The goal is not to invent new generation technology but to organize existing projects into a bankable package.

Once Firma identifies the right mix, it negotiates with the developers and presents the customer with a single contract rather than several messy agreements. That structure matters for both sides. For the data center, it reduces the administrative burden and creates a clearer path to clean power. For developers, it creates a buyer for projects that might otherwise languish without a customer. Firma says the approach can bring power online faster and at a competitive cost compared with gas plants. The claim is not that clean energy is always immediately available, but that the contracting and aggregation process can be compressed enough to make clean power a practical alternative.

Why gas still wins by default

The persistence of gas in AI data-center planning reflects a deeper set of incentives. Utilities and grid operators value dispatchable generation that can be called on when demand spikes. Natural gas plants can be built at a known scale and operated with familiar market rules. Their fuel supply is backed by infrastructure, and their permitting path, while not easy, is well trodden. In a region where a large data center is about to connect, a utility may prefer a gas plant because it can be planned within existing resource adequacy frameworks.

Clean-energy portfolios face a different set of hurdles. Interconnection queues are backlogged in many parts of the country. Transmission capacity may be insufficient to move power from windy and sunny regions to load centers. Permitting can take years, and local opposition can delay projects. Storage helps address variability, but batteries are still an added cost and require their own interconnection and safety reviews. A data center developer may conclude that waiting for a complex clean-energy package is riskier than signing a gas contract, even if the clean option looks attractive on paper.

Firma’s argument is that much of this friction is procedural rather than physical. If software can identify viable project combinations earlier, negotiate contracts in parallel, and give developers a clear signal that a buyer exists, then more clean projects can move forward. That could reduce the number of projects that stall for lack of a customer and make the clean-energy pipeline more responsive to large-load demand.

What a single contract changes for developers and buyers

A single contract may sound like a modest administrative improvement, but in power markets it can be significant. Large energy buyers often prefer simple, bankable agreements with clear pricing and performance terms. Developers, meanwhile, need certainty to raise financing and order equipment. When a data center tries to assemble clean power from multiple projects, each contract may have different terms, different curtailment rules, and different delivery points. That complexity can slow down legal review and make financing harder.

By aggregating projects, Firma aims to create a product that looks more like a conventional power purchase agreement. The customer gets a defined supply, and the developers get access to a buyer that understands the portfolio. The approach also allows for diversification. A mix of wind, solar, and storage can provide more consistent output than a single source, and it can reduce exposure to weather patterns or project-specific outages. For AI facilities that need high reliability, that diversification is a selling point.

The economics and limits of portfolio clean power

Firma says its approach can be competitive with gas on cost, but the comparison depends on location, grid rules, and contract length. In regions with strong wind or solar resources, clean generation can be cheap on a per-megawatt-hour basis. Adding storage raises the cost but improves firmness. Transmission constraints can erode the value of remote projects. Gas prices, meanwhile, can be volatile, and new gas plants face their own construction and fuel-supply risks. The real question is not whether clean power can be cheap in isolation, but whether a portfolio can deliver the right level of reliability at a price a data-center operator will accept.

There are also limits to what software can solve. A platform can match projects and accelerate contracting, but it cannot create transmission lines, clear a backlogged interconnection queue, or override local opposition. It cannot force a utility to change its planning assumptions. If the physical grid cannot deliver power to a data-center site, a clever contract will not fix that. Firma’s model therefore depends on picking regions where the grid can accommodate new load and where clean-energy projects are far enough along to be real options rather than speculative proposals.

Founders, funding, and early projects

Firma was founded by Jesse Jenkins, Scott Burger, and Brent Alderfer, energy experts with backgrounds in power systems, markets, and clean-energy development. The startup raised about $1.4 million in February 2026. It is now working on roughly a dozen projects with hyperscalers and AI cloud companies, according to the company. That early customer base is important because hyperscalers are among the largest and most demanding buyers of electricity. They have public clean-energy goals, but they also need power quickly and reliably. If Firma can satisfy both requirements, it could establish a repeatable model for other large-load customers.

The company’s focus on AI cloud companies makes sense. These firms are building the most power-intensive data centers, and they are often willing to sign long-term contracts to secure supply. They also face reputational and regulatory pressure to reduce emissions. A portfolio of wind, solar, and storage can help them meet clean-energy targets while adding new generation to the grid. The open question is whether the contracting model can scale from a dozen projects to the hundreds of projects that would be needed to serve the largest AI buildouts.

What the 200-gigawatt estimate really means

Firma estimates that just three grid regions could add 200 gigawatts of clean energy by 2030, a scale that would be large enough to support significant load growth if the projects can be commercialized. For context, 200 gigawatts is not a small number. It is comparable to a large share of the country’s existing generating capacity, and it would require enormous investment in wind, solar, storage, and transmission. The estimate should be read as a statement about potential, not a forecast. It depends on policy support, grid upgrades, permitting reform, and the willingness of buyers to sign long-term contracts.

The estimate also highlights why data centers are such a powerful force in clean-energy markets. A single buyer with a 500-megawatt load can anchor multiple projects. Several such buyers in the same region can create a pipeline large enough to justify new transmission and storage. That is the upside of the AI boom: it could accelerate clean-energy development by providing steady, creditworthy demand. The downside is that if the industry defaults to gas, it could lock in decades of emissions and make it harder to meet climate goals.

Grid, permitting, and hourly matching challenges

Even if Firma’s software works as advertised, the broader grid still has to absorb the new load. Data centers are often built in clusters, which can strain local substations and transmission lines. Utilities may need to upgrade infrastructure, and those upgrades can take years. Grid operators may require new resources to meet resource adequacy rules, not just annual energy matching. That means a portfolio of wind and solar may need storage or firm capacity to qualify. Some buyers are also moving toward hourly matching, where clean energy must be produced in the same hour it is consumed. That is a stricter standard than annual net-zero accounting and requires more storage or flexible demand.

Permitting remains another bottleneck. Wind and solar projects need land, local approvals, and environmental reviews. Transmission lines face even longer timelines. Battery storage is faster to build but still requires interconnection studies and safety approvals. Firma’s platform can speed up the commercial side of the equation, but it cannot eliminate these physical and regulatory constraints. The company’s success will depend on choosing projects and regions where those constraints are manageable.

A test for the next wave of AI infrastructure

The AI data-center boom is forcing the power sector to confront a question it has not faced in decades: how to add large amounts of new load quickly without abandoning clean-energy goals. Natural gas is the easy answer, and it will likely remain part of the mix for reliability and grid support. But Firma Power’s model suggests that clean energy can compete if the procurement process is redesigned around portfolios rather than single projects. By matching data centers with multiple wind, solar, and storage projects and packaging them into one contract, the startup hopes to turn a fragmented market into a faster, more predictable pathway.

Whether that model becomes standard will depend on execution. The company must prove that its software can identify viable projects, that its contracts can attract financing, and that the resulting power is reliable enough for AI facilities that cannot tolerate outages. It must also show that the approach works across different grid regions with different rules. If it does, the next wave of AI infrastructure could be powered by a more diverse clean-energy portfolio rather than a default dash to gas. The company’s early work with hyperscalers and AI cloud companies will show whether that model can move from a promising software layer to a standard way of powering the next generation of computing.

Source:Digital Trends News