Fiber Becomes Part of the Build
A decade ago, DC Blox might have prioritized land near existing fiber from AT&T, Verizon, Zayo or another established network provider. That calculation is different today. As data center campuses have grown and hyperscalers have become a larger share of the customer base, Wabik said connectivity has increasingly become another construction package associated with the project itself.
“If Zayo or Verizon or AT&T just happens to be there close by, that’s a good thing,” he said. “But fiber construction is inherently anymore just part of the construction component.”
That changes the site-selection question from Is there fiber nearby? to Can fiber be built here at the scale and diversity the customer requires? For DC Blox, Wabik said that can mean assessing whether sufficient public right-of-way exists to establish three or sometimes four diverse fiber paths into a data center.
That distinction is important, as AI workloads push infrastructure into markets where power, land and energy options may be more abundant than established carrier density. The hyperscalers themselves have also become major network builders.
Wabik characterized them provocatively as today’s telecom providers, pointing to the scale of terrestrial fiber they commission as well as the growing role of companies such as Amazon, Google and Meta in subsea cable development. The point is less that traditional carriers have disappeared than that hyperscalers increasingly design, commission and control enormous portions of the connectivity required to support their own infrastructure.
DC Blox now sees requests for 864-count fiber as routine and, in some cases, 1,728-count cable. That would have been difficult to imagine during an earlier era when a handful of fibers from an established carrier could satisfy a data center’s connectivity requirements.
AI-Scale Fiber Gets Physical
The scale becomes clearer when the discussion moves from abstract network capacity to what actually has to be put into the ground. Wabik said 864-count fiber is now about the smallest deployment DC Blox commonly sees. And the company is no longer designing conduit systems around the assumption that a modest amount of spare capacity will remain adequate for decades.
A few years ago, Wabik said, installing a handful of ducts during an initial build might have seemed sufficient for the next 20 years. Today, his minimum expectation is closer to 10 ducts, and sometimes 14. “I’ll put an 864 in it, I’ve got nine ducts left,” he said. “Who knows what’s coming and when it’s coming?” Sometimes the answer arrives almost immediately.
Wabik said a customer may request another 864-count cable before crews have completed the first installation. The second and third ducts can begin filling while the original construction effort is still underway. The density is equally striking inside hyperscale network facilities. Wabik described 10,000- to 20,000-square-foot network nodes containing hundreds of thousands of fiber splices, with equipment physically interconnected through an enormous web of fiber.
That scale mirrors what data center developers are seeing elsewhere in AI infrastructure. Wabik recalled hyperscaler requirements moving rapidly from 25 kW per cabinet to 35 kW and then 50 kW over a matter of weeks. The precise requirements can continue evolving while the underlying facility is being planned.
DC Blox’s job, he noted, is to provide the building, power, cooling and connectivity infrastructure around customer equipment whose final configuration may continue changing as deployment approaches. The same uncertainty encourages aggressive fiber over-provisioning. It is difficult to know exactly what the next requirement will be. The response is to create enough physical pathways to accommodate it.
“It’s just a massive buildout,” Wabik said. His shorthand for the moment is “beautiful insanity.”
Building a Connectivity Ecosystem
The result is something larger than a collection of point-to-point fiber routes. AI-scale campuses are becoming nodes inside increasingly dense connectivity ecosystems. DC Blox may construct fiber between hyperscale buildings, between campuses or into major network hubs. A customer can then ask the company to connect that newly built network with other fiber providers at multiple points along the route.
Those connections can require a series of intercepts and splices so the customer’s network can meet other networks at six or seven locations along a corridor. The picture that emerges is closer to a mesh than a traditional pair of routes entering a building. For hyperscalers, the ability to create that mesh is becoming part of the infrastructure equation from the beginning. But there is an important wrinkle in any discussion of fiber as a bottleneck.
Wabik does not see raw fiber deployment capacity as the obvious choke point. The industry is responding to demand by building at extraordinary scale. The harder problem often comes earlier. Before fiber can be placed, somebody has to grant permission to put it there.
The Hidden Constraint Is Permitting
When Data Center Frontier asked whether connectivity can keep pace when a developer finds large quantities of power, Wabik described data center development as a complicated choreography. Land has to be graded. Substations have to be built. Underground conduit has to be installed. Buildings have to rise. Metro and long-haul connectivity has to reach the campus. If any critical component misses the required date, the facility cannot operate as intended.
On the fiber side, Wabik said the hardest part is usually neither trenching nor horizontal boring. It is permitting. DC Blox has developed a roughly 475-mile route between Myrtle Beach, South Carolina, and Atlanta. Along a route of that length, fiber passes through numerous cities and counties, each potentially requiring its own equivalent of a building permit. Construction productivity can be measured. Crews can bore a certain number of feet per day and install a certain amount of fiber. A permit does not move according to the same clock. “Getting those permits is really the challenge,” Wabik said.
That has become important enough to affect routing decisions. If DC Blox knows a particular county previously took a year to approve a permit and became the long pole in a project schedule, Wabik said the company may simply route the next fiber build around that jurisdiction. That is a consequential development for an industry that usually discusses permitting primarily in connection with the data center campus itself. Permitting history can now influence the geography of the network connecting that campus.
A recent DC Blox metro Atlanta project illustrates the problem. Wabik said the company built roughly 55 miles of fiber to create a ring around the market. Most permits arrived during approximately the first two months of a 12-month construction window. Some did not. Several came in at what Wabik described as “right at the wire,” after months of follow-up with local jurisdictions.
Other infrastructure crossings make the process more complicated still. A bridge attachment can bring a highway department into the approval process. A river crossing may involve the U.S. Army Corps of Engineers. A railroad crossing introduces another third party with its own requirements and timetable. Wabik said DC Blox has seen individual segment permits take a year or, in some cases, 18 months.
The construction sequence then becomes an exercise in completing everything that can be built while waiting for the last approvals to land. Sending a crew home creates another problem. “If they have to go home, we have to remobilize them,” Wabik said. “That’s just a huge cost we really try to avoid.”
Fiber Supply Tightens Again
Permission to build is only one side of the problem. The physical supply chain has tightened again as well. Wabik said equipment lead times stretched dramatically during the COVID-era supply-chain disruptions, then eased during 2023 and 2024. Some fiber-related components returned to lead times measured in weeks.
That respite did not last. If DC Blox looks for 864-count or 1,728-count fiber today, Wabik said the lead time can reach 70 to 80 weeks when buying at what he considers normal market pricing. For an 864-count microfiber of approximately 10 or 11 millimeters in outside diameter, he said he would expect pricing around $7 to $7.50 per foot.
Inventory can be found more quickly for buyers willing to pay. Wabik said fiber sitting in available inventory may trade for $10, $12 or even $14 per foot when a project urgently needs material. The same pressure extends beyond connectivity. Large generators used for secondary data center power can also carry lead times in the 12- to 18-month range.
The challenge, therefore, is not simply obtaining one scarce component. Developers are coordinating a collection of long-lead systems whose delivery schedules have to converge on the same campus.
The Workforce Behind the AI Build
The physical scale of the build is also creating demand for people far beyond the AI software workforce that tends to dominate public discussion. Wabik pointed to real estate professionals locating land, specialists who understand how to work with utilities, construction crews grading enormous sites, plumbers, electricians, fiber technicians and the operating personnel required once hyperscale facilities come online.
Some large campuses can involve thousands of workers across the construction cycle. The constraint is increasingly visible in skilled trades. “There’s a shortage of electricians in the world right now,” Wabik said. Operations presents another challenge. Hyperscale facilities can contain hundreds of thousands of square feet of white space, requiring personnel capable of maintaining increasingly large and technically demanding environments.
In other words, the AI infrastructure build is creating a workforce requirement far broader than model development or semiconductor engineering. The servers only function after someone finds the land, secures the power, pours the foundations, installs the electrical infrastructure and connects the fiber.
Community Acceptance Enters the Critical Path
Another requirement has moved much earlier in the development cycle: community engagement. Wabik remembers an era when data center developers could enter communities with relatively little controversy, particularly when projects replaced distressed industrial assets.
A developer might acquire a former steel mill, remove an abandoned facility, remediate contaminated soil and introduce new investment. The current environment is markedly different. Community concerns around power consumption, water use, noise and environmental impacts have become central development issues in many markets.
Wabik argues that some of those concerns are driven by incorrect or overly broad assumptions about how individual facilities operate. He points to DC Blox’s own cooling architecture as one example. The company’s closed-loop systems may require roughly 500,000 gallons of water during commissioning, he said, with that water then recirculated rather than continuously consumed. He contrasted that design with facilities using evaporative cooling, which can have materially different water requirements.
The broader lesson for DC Blox has been that developers cannot assume communities understand those distinctions. The company is moving stakeholder engagement earlier and providing more detailed information about projects before construction. That can include expected sound levels, water consumption, environmental impacts, job creation and anticipated tax contributions.
Wabik noted that a large data center campus may represent several billion dollars of taxable property, along with business personal property inside the buildings. In some jurisdictions, a data center can become one of the largest individual taxpayers. Over a facility life that may extend 20, 40 or 50 years, he said the resulting tax contribution can reach hundreds of millions of dollars.
DC Blox increasingly sees communicating that complete project profile as part of development itself. That does not eliminate legitimate community questions about power, land, water or development policy. It does mean that the technical design of a project and the public understanding of that design can no longer be treated as separate exercises.
Everything Has to Arrive Together
The AI infrastructure boom has made power the dominant site-selection constraint for good reason. Without electricity, nothing else matters. But Wabik’s view from the fiber side shows why the next generation of development cannot be reduced to a single-variable search for megawatts.
Once the power is found, the network has to reach it. The right-of-way has to exist. The jurisdictions along the route have to approve construction. Fiber must be ordered far enough in advance. Crews have to be available when the permits arrive. Network paths have to multiply as customer requirements evolve. Communities have to understand what is being proposed before opposition hardens around assumptions that may or may not describe the actual project.
None of those requirements is particularly exotic on its own. What is new is their scale, their speed and the degree to which they are becoming interdependent. AI campuses are forcing developers to assemble power plants, substations, buildings, cooling systems, fiber networks and workforce ecosystems on timelines that increasingly overlap.
That may be the clearest meaning behind Wabik’s “beautiful insanity.” The fiber can follow the power. But power, connectivity, permits, equipment, labor and public acceptance increasingly have to arrive at nearly the same time.
