Nvidia Commits to 8,000-Mile Zayo Fiber Build as FERC Tariff Deadline Hits August 17

Zayo scales AI backbone with Nvidia as anchor customer

Zayo Group is building more than 8,000 miles of new long-haul fiber-optic cable across North America with Nvidia committed as the anchor customer. The expansion adds six brand-new long-distance routes and upgrades capacity in 10 high-demand markets, bringing Zayo's AI-focused network footprint to more than 15,000 route miles.

For enterprise buyers planning GPU-heavy workloads or multi-region AI deployments through 2027–2028, this is a concrete backbone project you can reference in RFPs. The new routes increase path diversity for private line, Ethernet, and wavelength services between AI data center clusters. Buyers should ask specifically for Nvidia-anchored routes or the upgraded markets when designing DC-to-DC and DC-to-cloud interconnects.

The anchor-customer model positions Zayo similarly to hyperscale-backed fiber projects where AWS, Microsoft, or Google commit to large backbone capacity. It reinforces Zayo's role as a go-to provider for GPU data center clusters and marginalizes smaller regional long-haul operators that lack route diversity across AI corridors or pre-engineered 400G/800G wavelengths.

What this changes for procurement and risk

The expansion increases path diversity, potentially improving resiliency for mission-critical long-haul traffic. However, concentration of AI traffic onto Zayo routes raises the importance of multi-provider strategies. Enterprises should consider dual-homing with Lumen or Verizon for risk mitigation, particularly if AI workloads represent core revenue or operational dependencies.

No public per-mile or per-wavelength pricing has been disclosed. Procurement teams should expect premium pricing for low-latency, AI-optimized routes and better leverage in negotiations if they can commit multi-year or multi-route contracts, especially in the 10 upgraded markets. Explicitly ask Zayo for SLAs covering latency, jitter, and repair times on the new AI corridors, and clarify any capacity reservation or prioritization clauses favoring Nvidia traffic.

Direct competitors include Lumen Technologies, Verizon, AT&T, and Cogent, all of which operate long-haul fiber suitable for AI and data center interconnect. The Nvidia commitment gives Zayo a clear differentiation point, but buyers should compare route-specific latency and diversity claims across providers before committing.

FERC interconnection tariff changes force August 17 filing deadline

The US Federal Energy Regulatory Commission issued a show-cause order stating that major grid operators' rules for large load interconnections—including data centers—are inadequate and must be revised. Regional operators including PJM, SPP, CAISO, ISO-NE, and NYISO face an August 17, 2026 deadline to file revised interconnection tariffs or justify why existing rules already comply.

The critical change: data centers must pay the full cost of grid upgrades directly tied to their own interconnection, rather than spreading those costs across all ratepayers. FERC identified five areas operators must address: efficient transmission application and study processes, prevention of cost shifts and transparency into transmission costs, rules for co-location and behind-the-meter generation, new transmission services for flexible large loads, and study processes for generating facilities serving electrically proximate large loads.

Co-location becomes more attractive as greenfield costs rise

The cost allocation directive makes co-location strategies from large cloud providers and wholesale operators—Equinix, Digital Realty, QTS, Vantage, Switch—more attractive relative to greenfield private enterprise builds. Big operators can amortize interconnection upgrade costs across large customer bases, while smaller enterprises attempting their own campuses will face higher, more transparent grid upgrade charges.

Behind-the-meter generation and on-site power strategies gain importance. Enterprises planning new data center capacity should model scenarios where grid interconnection costs exceed initial expectations by 20–40 percent, particularly in constrained markets like Northern Virginia, Phoenix, or Dallas. The tariff changes also increase the value of flexible load profiles that can shift consumption based on grid availability, though few enterprise workloads currently support this.

For enterprises with 2027–2028 data center expansion plans, the August 17 tariff filings will clarify the actual cost structure for new builds versus co-location. Budget holders should wait for the revised tariffs before finalizing site selection or procurement strategies, particularly if evaluating greenfield options in PJM or CAISO territories.

What to watch

Zayo has not disclosed pricing for the new AI routes or the 10 upgraded markets. Enterprises should ask for route-specific latency benchmarks, diversity claims, and SLA commitments during vendor discussions, particularly for paths connecting major AI regions. The absence of published pricing creates negotiation leverage for early commitments but also obscures true cost comparisons across Lumen, Verizon, and AT&T.

The August 17 FERC deadline will produce revised interconnection tariffs from every major US grid operator. These filings will set the baseline cost structure for new data center builds through at least 2028. Enterprises evaluating greenfield sites should delay final decisions until tariffs are published, or structure contracts with cost-adjustment clauses tied to final interconnection charges.

The Nvidia-Zayo partnership signals that hyperscale AI operators are locking in long-haul capacity years ahead of demand. Enterprises waiting for pricing to normalize may find themselves competing for scarce wavelengths or accepting higher latency routes. If your 2027–2028 roadmap includes GPU clusters in multiple regions, start conversations with Zayo, Lumen, and Verizon now to understand route availability and capacity reservation terms.