Minnesota Power's $450M HVDC Rebuild: Replacing Aging Infrastructure or Gold-Plating for Google?

Engineering News-Record reported that Minnesota Power has begun construction on a ^[1] $450 million high-voltage direct-current modernization project with Mortenson as general contractor, replacing transmission infrastructure nearly 50 years old. The 465-mile line connecting Center, North Dakota, to Hermantown near Duluth will receive new converter stations and substations, allowing bidirectional power flow. So far, standard utility work. But the timing and the driver demand scrutiny.

The project's catalyist is a proposed $2 billion Google data center planned for Hermantown. ^[1] This is where the mechanism shifts from "replacing worn equipment" to "whose cost, whose benefit." When a transmission project's primary demand comes from a single private corporate customer planning a massive new load, and the utility builds it through rate-base cost-plus financing (earning a FERC-regulated return on every dollar of capex), the structure inverts: Minnesota ratepayers bankroll the infrastructure, Google gets the juice, and Minnesota Power gets a guaranteed margin on the entire bill. The Google data center does not pay for the line; Minnesota's residential and business ratepayers do, in their monthly charges. Minnesota Power's CEO framed the modernization as pure grid benefit: "flexibility" and bidirectional flow during extreme weather. ^[1] That is a real operational gain. But it does not answer the cost-allocation question: if the line's core new demand is Google's data center, why are Minnesota ratepayers bearing the full capex cost rather than Google paying a connection charge proportional to its load?

This is not a novel problem, but it is structural. Transmission projects are evaluated and cost-allocated through regional planning processes (under FERC Order 1000 in most of the country) that theoretically sort "regional benefit" from "local benefit" and allocate costs accordingly. Minnesota Power operates in MISO, which adopted Order 1000's competitive-bidding framework in principle, though incumbents retain substantial planning discretion for local and reliability projects. A project framed as "asset replacement" (aging 1977 infrastructure) can clear with minimal interrogation of its true demand driver. If the line's actual capacity expansion is driven by Google's data-center workload, that demand should trigger a separate cost-allocation analysis, ideally one that names Google as a beneficiary and tests whether competitive bidding or a direct connection charge would be cheaper than incumbnet-utility cost-plus construction. Whether that analysis happened, and what it concluded, is not public in the available reporting.

The deeper issue is that data-center load is becoming transmission's fastest-growing hidden demand. Tech firms site facilities near existing major load centers and transmission hubs, then rely on utilities to upgrade the lines carrying power to them. Because the upgrades are built by the incumbent transmission owner and rate-based through tariff cost-allocation, the data-center customer pays only a fraction of the actual cost. Microsoft, Google, and others have explicitly lobbied for fast-tracked transmission permitting and cost-allocation rules that treat their loads as "grid benefits" rather than private demand. The Minnesota Power line may be the first $450 million articulation of that strategy in a single project announcement.

The questions to press now: First, did Minnesota Power perform an independent grid-enhancing technologies assessment, dynamic line ratings, advanced power-flow control, or reconductoring with advanced conductors, before committing to a $450 million rebuild? If the line's core function is moving Google's data-center power, could storage-as-transmission or demand-response engineering have deferred or reduced the capex? Second, what does the MISO cost-allocation determination say about who bears the expense? If Google's load triggered the upgrade, a portion should hit the customer's connection charge, not Minnesota ratepayers' baseline tariff. Third, is the $450 million estimate competitive? FERC Order 1000 opened large regional projects to competitive bidding precisely to strip out cost-plus padding; utility incumbents frequently bid against third-party transmission companies and lose or match on price discipline. Did Minnesota Power solicit a competitive bid, or was this sole-sourced to the utility's own capital program?

The fact that the line is genuinely aging and performing a real function, and that the HVDC technology upgrade is sensible, does not resolve the cost-allocation issue. Replacing a worn line is not the same as expanding capacity to serve new private demand. If Minnesota Power is doing both, ratepayers deserve to see the split and know why they are paying Google's grid access fee. The alternative is transparent: name Google as a beneficiary, model the line's capacity before and after, allocate the incremental capex to the incremental load, and either charge Google a proportional connection fee or open the project to competitive bidding to prove the $450 million number is real.