This week at the ASU+GSV conference—an enormous education innovation Summit held each year in Phoenix—Evan Marwell, founder of Education Superhighway, gave a lecture in a tiny crowded room with a low ceiling and not enough chairs to accommodate an eager audience. Before the presentation even started, the projector went out, so Marwell proceeded sans PowerPoint, reciting facts and figures about his organization’s efforts to measure the state of schools’ broadband and wireless access. When the projector screen eventually came to life, the resolution was a bit fuzzy. It seems to be Murphy’s law that when attempting to present on the power of education technology, the technology waivers. The irony of the flickering screen was not lost on Marwell, however, who spends his time explaining that for digital learning to really take hold we have to wake up to a startling reality: 72 percent of K–12 public schools in the U.S. do not have sufficient Internet infrastructure for digital learning, and that number stands to grow in the coming years as demand for more broadband continues to rise. In other words, the most basic fuel for the education system to harness the potential of education technology is not in place.

The ASU+GSV Education Innovation Summit is a see-and-be-seen of the edtech world, where entrepreneurs, heads of school systems, and big name investors come to learn about education product development and broker deals. As Chip Paucek, the co-founder and CEO of 2U, an online learning company that went public last month, put it, “You can see everything in the space in two-and-a-half days.”

But Marwell’s brief presentation has grave implications for the success of many of the businesses that flock to the Arizona desert each year to pitch and invest in educational technology systems. Software vendors are rapidly developing tools to help teachers and parents better personalize education through adaptive learning programs, gaming apps, and sophisticated assessment platforms. Yet many of the conversations among these start-ups and growth companies alike focus on the features and functionalities that their products can offer to schools and families. They are shockingly silent on the issue that Marwell illuminated in that tiny room, though.

Why aren’t edtech businesses more worried about broadband access? As Marwell explained, “Vendors are waiting for someone else to fix it.” Of course, from a business and cost perspective, no one individual educational software company is incentivized to take this on. The exorbitant costs of broadband would make subsidizing broadband unsustainable (Marwell thinks that reaching lower price points would actually require increasing competition among telecomm providers and a broader set of federal E-Rate policy reforms). Not to mention, any vendors trying to solve the broadband issue by themselves would also face a free rider problem, as others in the industry would stand to profit on the other end of this basic infrastructure investment.

But software vendors can still play a role in helping build awareness at the superintendent and district levels. As Marwell put it, “Districts don’t have to accept the state of their networks.” In his opinion, software vendors are well positioned to do two things.

First, vendors can raise awareness among their customers and invest time in finding ways to connect them to wireless and broadband solutions.  This means helping school systems find ways to support networking solutions beyond federal E-Rate subsidies. As Matthew Peskay, the chief of innovation and technology of KIPP LA Schools—which is pursuing a bold blended-learning model across the network—explained, “You need to be able to fund whatever your plan is without E-Rate funding.” Software vendors can help school systems grapple with these challenges, connecting them to services, like wireless providers, to strengthen their networks.

Second, software vendors can be transparent about the bandwidth that their applications use. Marwell explained that despite limited bandwidth in schools, most vendors don’t engineer products to minimize bandwidth requirements for their programs. Because nobody in the industry publishes this data in a standardized way, it can be difficult for schools to plan ahead for how their networks will be able to accommodate numerous software products working at once. Marwell imagines that down the line there might be labs to test hardware and software in order to measure how much bandwidth you will need for various programs to be up and running.

It’s in software vendors’ interest that their products are implemented in environments that lend themselves to success. Disruptive innovation theory would also posit that vendors who could get their broadband requirements down might successfully target pockets of nonconsumption and the low-end of the market, among schools whose current networks offer limited connectivity, but where teachers are still trying to integrate software into their classrooms.

The vibrant energy at this year’s ASU+GSV summit reflects the enormous growth potential of the edtech market and the unprecedented private capital flowing into software solutions for teachers and schools. But as online learning marches upmarket, we can’t ignore the basic unmet infrastructure needs inside the vast majority of America’s school buildings.