Emerald AI has raised $150 million in a Series A round at a valuation of $1.05 billion. The deal gives the company unicorn status, a term used for a private company worth more than $1 billion. The round was announced on August 25, 2026, and was co-led by Energize Capital and DCVC.
The company has a simple but important goal. It wants to help AI data centres use electricity in a smarter way. AI systems need huge amounts of power, and that need is set to grow as more companies build large AI models and services. Emerald AI says its software can help data centres change their power use when the electric grid faces high demand.
This idea puts Emerald AI at the centre of two fast-growing areas: artificial intelligence and energy. AI needs more data centres, while data centres need more electricity. The company believes software can help both sides work better together.
Why Data Centre Power Matters
Modern AI data centres can use a very large amount of electricity. Large AI systems need powerful computers, and those computers must run for long periods. As companies build more AI infrastructure, the demand for power also rises.
The problem is that electric grids cannot always add new capacity quickly. New power lines, substations and other grid projects can take years to build. Emerald AI says some new grid infrastructure can take a decade or more.
At the same time, the International Energy Agency expects data centres to account for nearly half of the growth in US electricity demand through 2030. That creates a serious challenge for utilities, data centre operators and local communities.
Emerald AI believes data centres should not always act as fixed power users. Instead, they can adjust their electricity use when the grid needs help.
How Emerald AI’s Software Works
The company’s main product is called Emerald Conductor. It is software that manages AI computer workloads and energy resources at a data centre.
When the power grid faces stress, the software can reduce the facility’s power use. It can do this while protecting key AI tasks that cannot stop. This means some computer work can slow down, pause or move to another time, while critical work can continue.
The idea is similar to turning down power use at the right moment rather than shutting down an entire data centre.
Emerald AI also works with power resources at the site, such as batteries and other energy systems. By coordinating computer workloads with available power, the company aims to give utilities more control over electricity demand.
This can help create a data centre that reacts to the needs of the grid instead of placing more pressure on it.
A Different Way to Build AI Infrastructure
The usual approach to AI data centres is simple: build the facility, connect it to the grid and secure enough electricity for its computers.
Emerald AI wants to change that model.
Its software can make a data centre more flexible. If the grid has plenty of power, the facility can use more electricity. If the grid has less available power, the facility can lower its demand.
This flexibility could have value for both sides. Data centre operators could gain access to power faster, while utilities could gain a large customer that can reduce demand at key times.
Emerald AI says this approach could unlock more than 100 gigawatts of unused capacity on the existing US grid for AI. The company says this could provide power years before some new grid projects are ready.
The Company Has Moved Beyond Tests
Emerald AI is not at the research stage alone. Over the past year, the company completed five commercial demonstrations at data centres in Arizona, Illinois, Virginia, Oregon and London.
The tests involved major companies and energy groups, such as NVIDIA, EPRI, Oracle, Nebius and National Grid, along with regional utilities and grid operators. These projects showed that AI data centres can change their power use based on grid needs.
The company has now moved into commercial scale. Its software is deployed at multi-megawatt data centre sites, and Emerald AI says it serves customers across the AI and energy sectors.
One full data centre in California has already shown that it can adjust its power use during periods of high grid stress.
That step matters because a system that works in a small test must also work at a much larger scale before major utilities and data centre operators can rely on it.
A Major Test in Virginia
One of Emerald AI’s most notable projects is in Manassas, Virginia.
The company is working with Digital Realty and NVIDIA on the nearly 100-megawatt Vera Rubin AI Research Factory. The site is designed to show how an AI facility can respond to power needs from the grid.
The project has also involved EPRI, Dominion and the PJM Interconnection. Emerald AI says the facility is set to come online later this year.
The project could become an important example of how large AI facilities and power grids can work together. If the model proves successful, it could support wider use of flexible power systems at other AI data centres.
Big Names Back the Company
The $150 million Series A has attracted a large group of investors from both the technology and energy sectors.
Energize Capital and DCVC co-led the round. Other investors include NVIDIA, Samsung Ventures, Siemens, Aramco Ventures, Salesforce Ventures, GE Vernova, RWE and JERA Ventures.
The list also includes In-Q-Tel, Radical Ventures, Energy Impact Partners, Lowercarbon Capital, Emerson Collective and several other financial and strategic investors. John Doerr and Tom Steyer also took part as individual investors.
Emerald AI says 12 Fortune Global 500 companies are now investors in the company. These companies also sit on its Strategic Advisory Board and work with Emerald AI on product use and commercial deployment.
The mix of investors is important. It shows that the power problem around AI has interest from both technology companies and energy companies.
What Emerald AI Plans to Do Next
The company plans to use the new capital to expand commercial deployments around the world.
Its customers include AI companies, data centre operators and electric utilities. Emerald AI wants its software to become part of the wider AI power system as more data centres come online.
The company has also worked with Silicon Valley Power on a Flexible Load Interconnection Program. Under this model, data centres can seek greater grid access in exchange for verified flexibility in their power use.
This type of deal could become more common if AI power demand keeps rising. Instead of asking only how much electricity a data centre needs, utilities could also ask how flexible that demand can be.
Why the $1.05 Billion Valuation Matters
The $1.05 billion valuation shows that investors see the power problem as a major part of the AI economy.
For years, much of the AI race focused on chips, models and computing power. Now electricity has become just as important. A company may have the best AI chips in the world, but those chips still need a reliable source of power.
Emerald AI is betting that software can help solve part of this problem.
Its success will depend on whether data centre operators trust the system, whether utilities accept flexible AI loads and whether the software can work at a very large scale. The company has already completed commercial tests, but the next stage will bring much larger deployments.
The Bigger Picture for AI and Energy
Emerald AI’s rise shows how closely AI and energy have become linked.
More AI means more computers. More computers mean more data centres. More data centres mean more demand for electricity. That demand can put pressure on power grids and local communities.
Emerald AI offers a different answer. Instead of treating data centres as fixed electricity users, it wants them to act as flexible customers that can change their power use when the grid needs relief.
The company has now raised $150 million at a $1.05 billion valuation, completed five commercial demonstrations and moved toward wider commercial use. Its next challenge is to prove that this model can work across a much larger part of the AI industry.
If it succeeds, Emerald AI may show that the future of AI data centres is not only about more power. It may also be about better control of the power they already have.
ALSO READ: WRTN Raises $72.2M in Series C