Instead, they observe that the emergence of AI is accelerating existing technology pressures districts have been grappling with for years: aging devices, an increasing dependency on cloud services, sophisticated cybersecurity threats and chronic staffing shortages. According to some experts, the central question for district leaders is not whether a network can run AI, but how well it can support districtwide, everyday use at scale over the next several years.
WHAT COUNTS AS AI INFRASTRUCTURE?
In 2025, the nonprofit Consortium for School Networking (CoSN) released a report stating that only 16 percent of K-12 schools at that point had reported being fully ready for AI with necessary tools and systems, and 61 percent had dirty and/or siloed data not ready to be operationalized by AI.
CoSN board member Tom Ryan, who is also a senior fellow of the Center for Digital Education and co-founder of the K-12 Strategic Technology Advisory Group, said AI readiness is about the foundational systems, policies, protocols and facilities that schools need to function effectively. These include infrastructure governance, future bandwidth forecasting and procurement protections.
Within the school building, Ryan said AI infrastructure will mean wireless connectivity with the capacity to handle “microbursts” — sudden online disruptions that occur when thousands of devices simultaneously log in or connect to cloud systems.
“We’ve got to provide bandwidth at a significant capacity ... but when everybody turns the computer on in the morning, you get these microbursts that shoot way up beyond your capacity to hook up to Microsoft, download the latest update or whatever there might be,” he said.
Hardware is another critical piece of the puzzle. Jason Neiffer, executive director of the Montana Digital Academy, a state-sponsored K-12 program, noted that while AI is currently cloud-based, the devices in students’ hands matter, too. He suggested that if technology trends see AI workloads eventually shifting back away from the cloud to local devices again, districts will save on cloud costs but need “more girthy hardware.” He questioned the longevity of low-cost devices, like Chromebooks with only two to four gigabytes of random access memory (RAM) that many districts relied on during and since the pandemic.
He argued those devices already provide poor user experiences, because students are discouraged when “school equipment is ... far from as functional as something that they may be using at home.”
AI ISN’T BREAKING NETWORKS ... YET
Despite concerns about AI’s impact on school networks, both Neiffer and Kris Hagel, chief information officer at Peninsula School District in Washington state, said they have not yet seen major AI-specific bandwidth crises like Wi-Fi congestion.
Neiffer instead described AI as one additional layer added onto an already expanding digital environment in schools, such as streaming, 1:1 device use and computer-based testing. However, he warned that the accumulation of these demands is relentless.
“I’ve heard a number of folks in the tech leadership space say things like, ‘however much bandwidth you have today, you know you will need 200 percent more in five years,’” he said. “It’s just not the way modern schools are set up.”
According to these technology leaders, one danger for districts is underestimating the complexity of the AI revolution. Cybersecurity, for example, is a primary area where AI is already increasing the load on school network capacity.
Ryan said school infrastructure must now include filters and access controls to prevent bad actors from using AI to steal data or spread fake or harmful content. He emphasized that districts must build guardrails for privacy, security and legal compliance before wide-scale deployment.
Moreover, Ryan repeatedly framed AI infrastructure as a strategic-planning issue rather than a simple technical benchmark. He said districts cannot wait until networks fail before upgrading systems, mainly because large-scale infrastructure projects — such as expanding wireless capacity — take years to plan and fund.
“It just becomes a very complex and hard-to-manage environment that needs to be well thought out and not reacted to,” Ryan said, adding that the real AI infrastructure challenge for schools is not simply “more bandwidth.” It is the transition from static, predictable school networks to dynamic, always-on, cloud-dependent ecosystems that require simultaneous investment in things like cybersecurity, data systems, staffing, teacher training and device capability.
THE FUNDING PROBLEM
Yet while the federal government pushes to expand AI in education, the move toward technologically fortified infrastructure is hampered by a nationwide financial squeeze.
“There’s nowhere near enough funding in K-12 education,” Hagel said, noting that many districts across the U.S. are facing funding crises due to enrollment declines. “I’ve never seen more school districts in the state of Washington with what’s called binding conditions, which is essentially … they’re bankrupt.”
He noted that the federal E-rate program, which provides annual broadband subsidies to schools and libraries, helps cover the cost of tech necessities like Wi-Fi, but it does not cover the operational costs of AI. According to Hagel, many districts are underestimating the long-term fiscal impact of AI, particularly the costs of cloud dependency and the need for infrastructure beyond simple bandwidth. He said schools are increasingly “paying essentially per-token costs to providers like Amazon Web Services or Microsoft Azure” — costs that could scale quickly as more students use the tools.
Neiffer also said K-12 districts have historically underinvested in professional learning, particularly regarding technology implementation. That will have to change if AI is to be expanded and effective in their learning environments.
Hagel echoed this sentiment, introducing the concept of “human bandwidth” as a component of AI infrastructure: As AI-integrated filters flag more student interactions for safety or mental health concerns, the burden for dealing with them falls on staff.
“There’s an increase in the human bandwidth that we’re going to need to triage some of those that might be more concerning,” he said.
Several experts said funding concerns are most visible in rural and underserved areas.
“The challenge we have isn’t that the schools don’t have bandwidth, especially because of E-rate … it’s that there’s not available bandwidth in the community,” Ryan said. For example, he pointed to remote regions like the Navajo Nation or southeast New Mexico, where Internet providers lack capacity, and some communities still even lack basic electricity.
In Montana, Neiffer said “bandwidth in every way, shape or form, is always a struggle with rural districts” and that a lack of internal tech departments or full-time CIOs remains a constant barrier to scaling technology.
