For the last several months, one question has come up in my conversations with operators more than any other. It isn’t whether AI data centers are real, or whether rural markets have a role to play. Most operators have moved past those questions.
The question is: How do I get into the conversation? It’s the right question. But most people are asking it too late.
What the Bandwidth Numbers Tell Us About AI Infrastructure
Zayo released its 2026 Bandwidth Report this summer, and it’s worth your time. It analyzed purchasing activity across nearly 6,000 customers during 2025 and found that demand for long-haul dark fiber doubled year over year. Metro dark fiber demand increased by up to 20 times in top AI-driven metros due to inference requirements, and wavelength capacity purchases increased 2.6X among data centers.
The shift in circuit size is just as telling. In 2024, 400G wavelengths accounted for the largest total volume of terabits purchased, outpacing 10G and 100G solutions. A single 400G wave carries the equivalent of forty 10G circuits. This isn’t simply an existing customer upgrading. It’s a different kind of customer.
Who is that customer? Hyperscalers and carriers account for 95% of all long-haul purchases, but AI providers and neoclouds are emerging as a powerful new category of bandwidth consumer.
Yes, Zayo sells fiber, so take the findings with a grain of salt. But the direction matches everything else we’re seeing.
The Data Center Interconnection Queue Isn’t Getting Shorter
We’ve covered the power constraint here before, so I’ll keep this brief. The problem is getting worse, not better.
PJM has trimmed its near-term load forecast because of stricter data center vetting. Of the 16 GW of data center capacity targeted for 2026, only 5 GW has entered active construction. And here in Texas, Governor Abbott directed the PUC and ERCOT in August to audit all data centers advancing through the interconnection process before any additional projects may move forward, noting roughly 474 gigawatts of requests to connect, about 90 percent of them data centers. ERCOT halted its “Batch Zero” study until the audit is complete.
When giant campuses stall, compute doesn’t stop. It moves into smaller, more distributed nodes, closer to the people and businesses that use it.
I’ve started calling this community compute. It’s not a data center dropped into a rural county. It’s compute in our community, serving the community, which is a fundamentally different proposition from a hyperscale campus. It’s also a proposition rural operators are unusually well suited to.
Distributed compute is also connectivity-hungry. Every node has to reach the others, the cloud, and the users it serves. That’s what that 20x metro number is really telling us: inference is moving closer to demand, and every node that moves needs a bigger pipe.
At the Table or On the Table?
Someone put it to me recently in a way I haven’t been able to shake: Are you trying to be at the table, or would you rather be on the table?
Think about how a data center actually gets built. Site selection, power negotiations, land, financing and connectivity planning happen long before anything visible occurs. A site can be power-rich while remaining connectivity-thin, and fixing that second problem after construction begins may become increasingly difficult. So developers solve connectivity early, with the people who are already in the room.
By the time a local operator sees a permit filing or a groundbreaking, most of the vendor decisions have already been made.
Being at the table means trying to get a seat at someone else’s deal. You’re hoping the incumbent misses a date or the diverse route falls through. That’s not a strategy. That’s waiting for somebody else to screw up.
Being on the table means you’re what gets put forward: the site, the power, the path, the local relationships. You aren’t responding to the deal. You’re part of what makes the deal possible.
That changes everything downstream. You aren’t waiting months or years for someone to find you. You bring the opportunity to your community on your terms. The 100G and 400G services stop being something you hope to sell. They become part of the load you’re bringing to the table.
You’ve probably heard the older version of this line: if you’re not at the table, you’re on the menu. It applies here too. If a compute provider comes into your market and wins your best customers without you, you weren’t on the table. You were on the menu.
What Rural Broadband Operators Bring to Community Compute
None of that is possible until you understand what you have. That starts with three inventories:
- Your assets: Facilities with power and cooling already in place. Land. Your relationship with the local electric cooperative or utility, and whether there’s headroom at a nearby substation. Fiber routes, conduit, rights-of-way, and diverse paths.
- Your market: What’s moving around you? Planned developments, adjacent operators, and your routes to the nearest major interconnection point. Who is already looking at your region, whether you know it or not?
- Your customers: This is the one almost everyone misses.
Say you provide connectivity to the local hospital. That hospital is exactly the kind of customer a compute company wants to win. You already have the relationship, the trust, and the connection. A compute provider coming into your market would have to build all three from scratch.
That’s the real value. And you should expect to get paid for it, rather than give it away as the price of admission to someone else’s project.
Now multiply that across your enterprise base: the school district, the county, the ag processor, the regional bank. Your customer list may be the most valuable thing you bring to a community compute conversation, and it doesn’t show up on your balance sheet.
There’s one more asset rural operators undervalue right now. As communities push back on large data center projects, a locally owned operator with decades of standing in the community has something significant to bring to the table. Community compute only works if the community actually wants it, and you’re the one they already trust.
What you do with those inventories depends on your market, your balance sheet and your appetite for risk. But you can’t make any of those decisions until you’ve done them.
The Bottom Line: From Rural Broadband Provider to Compute Partner
The rural operators who participate in this buildout won’t be the ones who got better at getting invited. They’ll be the ones who stopped waiting for an invitation and did the work to become part of the offer.
So here’s the question I’d put to your leadership team: if a compute company called tomorrow and asked what you could bring to your community, could you answer in one conversation? Or would it take six weeks to find out?