For years, demand for solid-state transformers was largely theoretical. But several pilots — the latest announced this week — are beginning to demonstrate that there’s a real market: years from maturity, but concrete nonetheless.
DG Matrix and TerraFlow Energy are preparing to deploy what will be one of the first SST architectures to power a high-performance computing cluster in the U.S. The pilot project, part of a commercial agreement the companies announced today, will pair one of DG Matrix’s SSTs with one of TerraFlow’s vanadium redox flow batteries to power Dell servers.
An SST is a piece of equipment that converts electricity’s voltage and form using computer-controlled power electronics, which allows for more dynamic and streamlined control compared to a traditional transformer. As such, it has the potential to control the power spikes that are typical of AI data centers, especially when paired with energy storage.
In a recent equity research note, Jefferies analysts wrote that the SST market for U.S. data centers could reach $4.3 billion in 2030, up from $37 million today. Driving the growth is both hyperscalers’ urgency for power, which is elevating “power electronics from cost center to revenue enabler,” as well as long lead times for traditional transformers.
Haroon Inam, DG Matrix CEO, told Latitude Media the pilot project is a step toward meeting that growth. “It validates that not only is this a technology that people write papers about,… [but that] the economics are there, the scalability is there, the reliability is there,” Inam said.
DG Matrix already has 11 SST units active across the U.S., but this would be the first one to power actual high-performance computing, specifically a 1,200-core GPU cluster of Dell servers. “We’ve done the equivalent with simulated loads for almost a year and a half now, but this would be with a real load,” Inam said. “And that’s the big advantage, that people can come see it deployed with [TerraFlow’s] batteries.”
He added that the pilot will replicate Nvidia’s 800 VDC power architecture, which is becoming the chipmaker’s standardized basis for new AI data centers.
As Jon Parrella, TerraFlow CEO, explained, the idea is that the long-duration batteries combined with the SST will absorb the volatility of this load. “It’s like a marriage made in heaven,” Parrella said, “because now you have a battery that’s long-duration and can handle the pulse load or the volatility without degrading or catching fire.”
TerraFlow also plans on deploying its batteries with DG Matrix’s SST at its “Center of Excellence,” a 25-megawatt data center that will be a larger test case for the technology..
The ‘qualification window’
This project is just one of a series of pilots underway — including from companies such as GE Vernova, Amperesand, and Enphase — that are slated to determine not only whether the SST market takes off in the coming years, but which manufacturers will lead it.
For buyers, pilots represent one of the “credibility gates for vendor selection,” according to Jefferies, alongside manufacturing readiness. Analysts believe SST adoption at scale will start happening between 2027 and 2029, which means that between now and 2028 there’s “a qualification window” where vendors need to show that they have achieved results in the field. Jefferies describes DG Matrix as “one of the most commercially advanced SST startups,” especially for AI data centers. “Certification and field performance remain the critical proof points,” the note added.
If those deployments in the field are a success, hyperscalers are ready and waiting to buy. In fact, it’s their demand that’s propelling the SST market, as they look for solutions to simplify and accelerate their access to power — and are prepared to buy in large enough quantities to push SST developers towards commercialization.
“What AI data centers are doing is creating concentrated demand in the hands of a few hundred people, who are making large purchasing decisions,” Inam said. “The capital starts flowing in from larger investors to take down the cost curve in a record time.”
Andy Newbold is the head of corporate affairs at Enphase, which has just started manufacturing the first power modules for its SST products, scheduled to be piloted in early 2027. “SSTs aren’t necessarily new. Engineers have been working on them for 20 years, but on the traditional grid they always lost to the iron-and-copper transformer, which is cheap, passive, and lasts decades,” Newbold said in an email.
For years, he added, “the SST was an answer to a question nobody was asking.” Now, though, “AI data centers are asking.”
Early adopters
At the moment, SST prices vary from $30 cents to $1 per watt, which is the equivalent of $300,000 to $1 million per megawatt, according to Jefferies —- considerably pricier than traditional transformers. That said, SSTs replace multiple pieces of equipment, including both transformers and UPS systems, which for many potential customers will likely justify this higher sticker price. As Inam said, “We take out 11 to 18 different boxes of equipment and replace it with one.”
And regardless, Jefferies analysts added that the hardware cost has become less important for customers willing to do whatever they can to secure reliable power fast. In fact, the AI boom and the rush to secure power have changed companies’ attitudes toward technology risk.
Parrella says that’s evident not only with SST, but with the flow batteries his company manufactures, which also haven’t been fully commercialized yet. “Newer forms of battery technology are being tried…and accelerated, even if they’re not necessarily bankable yet, just because we solve more of the problem for data centers, so they’re willing to take a risk where they may not have in other situations,” he said.
While more conservative customers might be more cautious about new technologies, be they batteries or SSTs, TerraFlow is willing to nudge them toward adoption. The company is planning to embed DG Matrix’s SSTs in all its flow batteries, so customers can see the benefits for themselves.
“The early adopters are placing orders first, but once you see the deployments running in the field, that’s when we predict the next very large set of orders maturing,” Inam said. He added that DG Matrix is looking at closing between $700 million and $900 million in orders in the next few months, much of it for deployment in 2027 and 2028.
