OpenGrid will connect existing commercial and open-source modeling tools to make high-quality electrical grid planning tools accessible worldwide by 2030.

PRAGUE — AI companies have a problem, and it has nothing to do with guardrails and safety. As OpenAI describes it, “Electrons are the new oil, and we believe the Trump Administration should work with the private sector on an ambitious national project to build 100 gigawatts (GW) a year of new energy capacity.” That’s just for OpenAI. That begs the question: How do you build it and how do you manage it using the fragmented software mess used to plan and deploy it?

Open source is the answer. As Shuli Goodman, the late former executive director of LF Energy, said in 2020, the “long-term solution would be to open-source the entire stack. In essence, this means outlawing black boxes on the grid … And because any device attached to the grid can be a security vulnerability, we need a path toward complete transparency. A wiser course would be to recognize that any hardware with embedded, proprietary software that cannot be accessed or reviewed is the real threat to the grid.”

AI’s crushing demand for ever more power has led LF Energy to a new plan: OpenGrid. Its goal is to build a shared data and interoperability layer connecting existing commercial and open-source grid-planning tools. Rather than develop one replacement application, the initiative aims to make different modeling systems work together—and make their results easier for utilities, regulators and communities to inspect. This isn’t just for countries with energy-hungry AI users; it’s for every country that needs high-quality grid-planning tools.

It’s not just AI. Alice Yake, vice president of Breakthrough Energy GRIDS and chair of the OpenGrid Management Committee, in her keynote remarks at Open Source Summit Europe, said, “We’re going to see more than 5,400 terawatt hours of need come to the system between now and 2030. That’s more than a 20% increase in total energy consumption globally.”

“We don’t need to reinvent capacity expansion modeling, production cost, or power flow,” Yake continued. “What we need to do is to make them work in harmony and make these tools accessible so that regulators, decision makers, and people that are building these plans out for our communities can have transparency, repeatability, and ultimately alignment.”

Besides keeping AI vendors happy, she added, “Reliable electricity supports healthier lives, stronger communities and economic opportunity, but delivering it requires credible plans. Today, too many plans rely on data others cannot verify and tools much of the world cannot afford.”

The problem, Yake argued, is not a shortage of open-source energy-modeling software. We have lots of those. But using them is another matter. They often require specialized coding skills or advanced technical expertise. What has been missing is a shared responsibility for the broader ecosystem to work together.

OpenGrid will address that gap through open standards and schemas, programmatic APIs, integrations, and a centralized Data Hub containing validated, model-ready datasets. Its first flagship application, OpenGrid Translator, is in development for a 2027 release. The translator is intended to move data and models between widely used commercial platforms and open-source tools.

Eventually, OpenGrid plans to add open solvers, no-code graphical interfaces, standardized planning workflows and AI integrations to support data preparation. Those are roadmap items, not capabilities the launch announcement establishes as already available.

At the same time, electrical planners must account for variable generation, weather-dependent resources, and power moving in more than one direction through the system. This is not your grandpa’s coal-powered electrical grid. Yake said there’s a widening need to match new loads with the generation and network capacity required to serve them.

In the United States alone, she said, more than 166 gigawatts of infrastructure is seeking connectivity by 2030, much of it driven by data centers. Those requests don’t align with available generation capacity. Even setting aside the fact that many people hate the idea of data centers being built in their neighborhood and the resulting increase in their electrical bills, we simply don’t have the manufacturing resources to build new power plants. Even in 2020, Goodman noted, “The US has lost almost all capacity to build large high-voltage equipment, like transformers.”

Besides, even though the US now embraces nuclear energy, even the Department of Energy admits that it will take until 2030 to add just 5 GW of domestic nuclear energy to our power supply. Simultaneously, the Trump administration is actively working against green energy sources such as solar and wind power. We simply won’t have enough power.

That isn’t a problem OpenGrid can solve. OpenGrid’s narrower proposition is that better-connected tools and more accessible data can help planners evaluate and plan the infrastructure needed to address these needs.

OpenGrid will also, Yake remarked, give power companies and users access to high-quality planning tools that most companies can’t afford. OpenGrid will lower those barriers while making the technical basis of infrastructure decisions more transparent. Regulators, advocates, and other stakeholders can inspect the models and data informing proposed investments rather than rely solely on results produced inside inaccessible systems. As Yake said, “OpenGrid changes that: open data, open standards and access to planning tools that any utility, regulator or country can inspect and use.”

Still, we’re a long way from getting OpenGrid into the hands of the people who need it. Yake said the team had spent more than a year gathering feedback from almost 100 people across 13 countries, including regulators, operators and planners. The first concrete delivery milestone is OpenGrid Translator in 2027. The larger ambition, worldwide access to high-quality planning tools by 2030, depends on turning shared standards, datasets, and integrations into usable software.

We need OpenGrid, and we need it as soon as possible, no matter what happens with AI data centers, power plants, and grid buildouts.