Delivering the AI Era: Building a 100-Terabit Midwest Connectivity Corridor
An ISOC LIVE Summary
Delivering the AI Era: Building a 100-Terabit Midwest Connectivity Corridor
NEDAS Live Podcast Episode 66 – June 2026
Speakers: Priscilla Favors - Aureon; Chris Crowe - T3 Broadband; Kurt Raaflaub - Nokia; Jeff Sanders - Midco
Host: Ilissa Miller - NEDAS / iMiller Public Relations
This episode of NEDAS Live explored one of the largest coordinated wholesale transport initiatives currently supporting the AI infrastructure boom in the United States. Branded as a 100-terabit Midwest connectivity corridor, the project was built to support hyperscale data center growth, AI compute expansion, and the increasingly critical need for resilient, high-capacity transport infrastructure.
Ilissa Miller framed the conversation around a core industry reality: AI infrastructure discussions often focus on compute and data centers, but none of these systems function without the underlying transport networks moving massive amounts of data between facilities. She emphasized that the project represented one of the industry’s largest coordinated infrastructure collaborations and highlighted how rapidly AI requirements are transforming network planning and deployment.
Priscilla Favors explained that Aureon recognized early that AI, cloud infrastructure, and data center growth would dramatically increase demand across Midwest markets. She noted that historically, hyperscale infrastructure focused on major metropolitan areas such as Chicago and Dallas, but the search for available land and electrical power has shifted attention toward secondary and rural markets. She described the Midwest as being “at the right place at the right time,” where regional carriers could provide critical connectivity diversity and localized infrastructure support.
Chris Crowe described the broader technological moment as comparable to the Industrial Revolution, the computer revolution, and the rise of the Internet itself. He argued that AI now acts as the connective layer integrating all previous technological revolutions. According to Crowe, the hyperscale expansion underway requires a “nervous system” connecting distributed data centers, and that optical transport providers are effectively building the highways that allow AI systems to function together at scale.
The panel repeatedly stressed that the infrastructure challenge extends beyond simply constructing data centers. Jeff Sanders explained that the project required unprecedented coordination among transport providers, equipment vendors, construction crews, interconnection partners, and financial stakeholders. He compared the effort to a symphony in which every participant had to operate synchronously in order to avoid failure. Sanders emphasized that no single organization could have completed the deployment independently, making collaboration and trust fundamental to success.
Priscilla Favors added that many of the relationships underpinning the project had been built over nearly two decades in the telecommunications industry. She noted that hyperscale customers were simultaneously approaching multiple providers for solutions, creating intense competitive pressure and requiring rapid decision-making. Internally, Aureon had to accelerate financial approvals and technical coordination at unusual speed, including an emergency evening board meeting to secure capital commitments.
A major theme of the discussion involved supply chain constraints driven by hyperscale AI investment. Chris Crowe described how global shortages of components, cabling, amplifiers, and optical hardware complicated deployment timelines. He said that hyperscale spending has created allocation pressures across the entire telecom manufacturing ecosystem, with providers competing for access to essential equipment, and pointed to rising memory prices as one visible symptom.
Kurt Raaflaub expanded on this by noting that data center capital expenditures grew by more than 60 percent during 2025, exceeding $400 billion. He suggested that spending could reach a trillion dollars by the mid-2030s. Raaflaub stressed that regional carriers such as Aureon and Midco have become strategically important because AI infrastructure is increasingly decentralized away from traditional urban hubs. He argued that partnerships between regional operators, integrators such as T3 Broadband, and vendors such as Nokia are now essential to securing supply chain access and scaling infrastructure quickly enough to meet AI demand.
The conversation also examined how AI traffic is changing the technical requirements of optical networking itself. Kurt Raaflaub explained that the industry has moved beyond simply maximizing speed. Instead, operators are increasingly focused on spectral efficiency and power efficiency. He noted that fiber capacity is not infinite and referenced Shannon’s Law to explain the physical limitations of optical transport. According to Raaflaub, AI data centers consume such enormous amounts of power that reducing network energy consumption has become a critical economic consideration. He argued that older optical systems designed a decade earlier are no longer adequate for the scale and efficiency demands of modern AI infrastructure.
Jeff Sanders described how Midco has proactively upgraded its infrastructure to support emerging requirements including 400-gigabit and 800-gigabit wavelength services, managed optical fiber networks, and advanced DWDM capabilities. He emphasized that customer requirements now vary widely depending on geography, timing, redundancy, and capacity needs, making flexibility and adaptability critical operational priorities.
The panelists repeatedly returned to the idea that AI deployment remains only at its earliest stages. Ilissa Miller observed that the industry is still constructing foundational AI “factories” for large language models while simultaneously preparing for inference workloads, distributed AI systems, and future generations of agentic AI. She emphasized that future bottlenecks may emerge not only in compute, but also in transport and electrical power infrastructure.
Looking ahead five years, the panel predicted explosive continued growth in both network and power requirements. Jeff Sanders argued that the industry still underestimates the amount of intra-data-center fiber connectivity required inside AI campuses, noting that single AI facilities may consume more fiber than entire metropolitan deployments did historically. He also warned that power demand may evolve from today’s hundreds-of-megawatt campuses toward multi-gigawatt AI clusters.
Kurt Raaflaub introduced NVIDIA CEO Jensen Huang’s concept of the “AI grid,” describing a future where distributed compute becomes increasingly decentralized and pushed closer to users. He argued that optical networks will need to become more agile, lower latency, and more adaptive to support real-time AI interactions and distributed AI processing.
Chris Crowe predicted that future AI applications involving distributed video processing, complex datasets, and computer-to-computer collaboration would dramatically increase bandwidth demand beyond anything currently anticipated. He said that existing estimates of required connectivity are likely still too conservative.
Priscilla Favors highlighted additional emerging constraints around land acquisition, electrical power availability, and operational reliability. She explained that hyperscale operators increasingly face local resistance to data center construction and are sometimes unable to secure adequate power despite finding suitable land. She also stressed the enormous financial consequences of outages, citing one customer who estimated that a single day of downtime could cost $2.5 million.
In the closing discussion, the panel reflected on what the industry must prepare for next. Chris Crowe described data as “the new currency” and argued that communications providers are now building the highways for the future digital economy. Kurt Raaflaub stressed the importance of ecosystem partnerships between equipment vendors, integrators, and regional service providers in enabling ever-larger AI transport deployments.
Jeff Sanders concluded by warning that social and political resistance to AI data center development may become one of the industry’s defining constraints. Pointing to states such as Maine, where opposition has effectively blocked projects, he argued that telecommunications providers, hyperscalers, and infrastructure operators will need to engage more directly with communities and policymakers to cut through misinformation and address concerns around zoning, energy use, and local impacts.
The discussion ultimately portrayed AI infrastructure as a rapidly accelerating ecosystem challenge involving transport, power, supply chains, regional collaboration, optical innovation, and public policy. The panelists agreed that the industry is only at the beginning of a multi-decade transformation that will fundamentally reshape communications infrastructure worldwide.
RESOURCES
NEDAS Live! E66 — episode page for this discussion, hosted by Ilissa Miller
Aureon — Iowa-based wholesale fiber provider; Priscilla Favors’ company
T3 Broadband — Kansas-based optical network integrator; Chris Crowe’s company
Nokia — optical networking and data center interconnect equipment; Kurt Raaflaub’s company
Midco — regional fiber and connectivity provider; Jeff Sanders’ company
iMiller Public Relations (iMPR) — organizer of NEDAS Live!
Midco / Switch 400G deal — the historic multistate connectivity project referenced in the episode
What is a pluggable? — Kurt Raaflaub on the coherent optics driving the AI supercycle
800G ZR/ZR+ — Nokia on next-generation AI-scale optical connectivity
Priscilla Favors — LinkedIn, Aureon
Kurt Raaflaub — LinkedIn, Nokia
Jeff Sanders — LinkedIn, Midco