In Bryan, Austin-based ThisWay Global is building a 73-megawatt data center on 25 acres of Texas A&M's RELLIS Campus. The company calls the facility the Research and AI Compute Center, or RACC, and says it will focus on sovereign artificial intelligence — systems that let organizations keep control of their data and the information AI systems produce.

"Sovereign AI allows organizations to maintain greater control over their data and the information produced by AI systems," ThisWay Global CEO Stefan Fable said. "You have full control and you retain control both over the input as well as the output. That is extraordinarily important for research. It is extraordinarily important for regulated industry, for defense, for our mission that we are behind."

The center is also expected to include an educational component, with an eight-week course available to eligible students and community members. Fable said the goal is to provide training "across the entire spectrum of the professional, vocational and academic spectrum," extending opportunities beyond Texas A&M students to the broader community.

For students and researchers, access to computing power can be a significant hurdle. Texas A&M computer science Ph.D. student Sean Feeney said researchers often have to partner with industry to use larger compute systems.

"Having that closer to campus and more accessible alongside some of the other computing resources we have here, it's always a good thing," Feeney said. "To kind of keep the research and the compute local."

Arnold Castro, a professor of practice who teaches AI at Texas A&M, believes the center could give students and researchers another reason to remain on campus rather than leave in search of greater resources.

"It's going to allow our research to go to new levels," Castro said. He added that RACC could also provide a place for students to learn how to operate an AI data center as such facilities become more common.

The company plans to begin computing at the Bryan facility by June 2027.

In Dallas, technology integrator GDT has created what it calls an "AI factory," a supercomputer lab where businesses can develop and test artificial intelligence applications before investing in their own systems. From the outside, the facility may look like a glass room filled with cables, cords and racks, but GDT executives describe it as a new-age factory built around the computing infrastructure needed to run AI applications.

"We're now in the phase of where large enterprises are using AI to do work, and they have to deploy that," GDT CEO Shawn O'Grady said. "And that's a lot more complex than just searching ChatGPT on your iPhone."

The GDT lab uses NVIDIA GPUs, NetApp storage and Cisco networking. Companies can use the facility to determine whether an AI application works for their business before committing to the technology and infrastructure themselves.

Chris Kapusta, GDT's vice president of advisory and transformation, said the idea is to help companies move more quickly from exploring AI to determining whether a particular use has real value.

"For organizations that come in and use our AI factory, what it is, is it's really a way to accelerate that," Kapusta said. "How can we test out their use cases? How can we test out what they want to do? Make sure it makes sense for their organization, for their business."

Kapusta said having the facility in Dallas also gives North Texas businesses local access to the technology while potentially drawing national attention to the region.

"It allows a lot of Dallas local businesses to take advantage of what we're building here," Kapusta said. "We want to be a partner and help on this journey, but it also gives some national exposure because AI factories are kind of a hot topic right now."

The growth of AI and the infrastructure supporting it has not come without concerns, including questions about security and how the technology may be used. O'Grady compared artificial intelligence to any other tool whose impact depends on the person using it.

"AI is not a bad thing in the way that a hammer is not bad. It's a tool," O'Grady said. "And it's always in how the tool gets used. And you use a hammer to build a house; you can use a hammer to harm somebody. AI, the tool by itself, is not the problem."

Cybersecurity is one area where the technology can work both ways. Kapusta said bad actors are already using AI to search for security vulnerabilities and digitally infiltrate organizations. At the same time, he said companies can use artificial intelligence to find and address weaknesses before they become widespread.

"We can also use AI in a good way, so to speak, to combat those bad actors, to find and fix vulnerabilities before they become widespread," Kapusta said.

For those concerned about where the technology is headed, Fable said AI will increasingly become integrated into systems people already use, including payment systems, points of sale and other everyday technology.

"AI will be part of everything that you're already doing," Fable said.

While the Dallas and Bryan projects serve different purposes, both are designed around a similar idea: giving more people access to the computing resources needed to develop, test and understand artificial intelligence as its role in business, research and everyday life continues to grow.