The 7,235 MW interconnection request in Montana is not a data center story
Seven gigawatts, no tenant.
Related reading: NextEra’s reported bid for Dominion on the interconnection queue as the asset, and the Entergy/Meta Hyperion writeup on stranded-cost ratepayer risk.
A developer in rural Montana has asked to connect 7,235 MW to a utility whose entire system averages about 760 MW. It has disclosed a campus that ramps to 1,000 MW by 2031 and named no customer for any of it. The New York Times covered this as a transparency fight between a rancher and an opaque tech project. The filing describes something else.
The governing point is a category error. Broadview is being read as a data center story, and it is not one yet. What Quantica Infrastructure filed with NorthWestern Energy on 20 May 2026 is an interconnection position, six to seven times the size of the campus it sits on, assembled by an energy private equity sponsor that has been buying its way toward the relevant substation since 2020. The asset is the queue slot and the powered land underneath it. The data center is the use case that may or may not arrive to fill it.
The scale is what forces the question. NorthWestern runs about 760 MW across Montana on an average day and peaks near 1,300 MW, per its 2024 annual report. A single developer asking for 7,235 MW is asking for roughly 10 times the load of the utility it wants to plug into. The July 2025 letter of intent started at up to 1,000 MW. By Q1 2026 that had become a signed development agreement, which NorthWestern flagged in its first-quarter earnings highlights. Then in May the interconnection request arrived at 7,235 MW. The disclosed campus did not grow with it. Quantica still describes Big Sky as a roughly 1,100 MW campus, with Phase 1 ramping to 500 MW from 2028 and Phase 2 adding 500 MW from 2031. The customer count across that entire climb is still zero.
What has emerged in response is the apparatus you would expect when a multi-gigawatt speculative load lands on a small utility: a tariff fight. NorthWestern filed a Large New Load Tariff with the Montana Public Service Commission on 31 March 2026, built explicitly so that large new loads “pay their fair share” and existing customers “do not subsidise” them. Montana law already bars the utility from serving a 5 MW or larger load until the Commission finds it won’t harm other customers over the long term. The unresolved tension this piece traces is the gap between that protection on paper and what actually happens when the wire gets built before the tenant signs.
The asset is the queue position, not the campus
Read Quantica’s own language. The interconnection applications cover up to 7,235 MW of maximum more capacity, a mix of renewable generation, firming generation and battery storage, and they are “designed to support potential long-term customer demand.” That is not a description of a load. It is a description of an option on a load. The company has said the filing doesn’t mean the campus will use all the power, only that it is planning ahead.
You do not request an interconnection study for 7,235 MW against a 1,000 MW disclosed ramp because you have customers for the difference. You do it because the queue position itself is scarce and valuable, and because the formal review, as Quantica notes, has to begin years ahead of construction. The slot is the thing being warehoused. In a market where advanced interconnection position is the binding constraint on the AI power buildout, holding a multi-gigawatt study at a real substation is an asset you can monetise later by attaching a tenant, not a cost you incur because a tenant is already there. I made the same argument about Dominion’s queue in the NextEra writeup: delivery-point requests are not signed load, and the gap between the two is where the risk lives.
This is the financing-structure-misread-as-operating-signal pattern in its land-and-grid form. The press is reading the 7,235 MW number as a demand signal. The filing reads as an optionality position priced on the possibility of demand.
The capital tells you what this is
EnCap Investments is not an infrastructure fund. Since 1988 it has been a provider of growth capital to the independent US energy sector, with about US$47 billion raised across 25 funds. That is energy private equity: it underwrites development-stage upside, not stabilised contracted cash flow. No core infrastructure allocator underwrites an uncontracted seven-gigawatt queue position and books it as infrastructure. So the right question is not “is this a credible data center,” it is “what does an energy PE sponsor do with a powered-land position,” and the answer is develop it to a monetisable milestone and sell or lease the optionality.
The assembled history supports that read more than the campus narrative does. Quantica’s renewable assets trace back to Broad Reach Power, an EnCap-founded platform. Buffalo Trail Wind and Solar, near Broadview, was bought from Invenergy in 2020, and EnCap entities bought land around the Broadview substation that same year for what was then described as an impending connection. Broadview Solar, an 80 MW project at the same site, was the subject of a multi-year PURPA classification fight that NorthWestern waged and lost twice in federal court before declining to appeal again, around the time it signed the Quantica letter of intent. The substation position predates the AI framing by half a decade. The data center is the newest wrapper on a power-and-land assemblage that has been under construction since well before anyone was talking about gigawatt campuses in Montana.
Quantica’s CEO, John Chesser, the former CFO of Talen Energy, describes the product as land, power and high-speed fibre in one integrated solution. That is a landlord’s pitch, and it is an accurate one. The company calls itself a developer of integrated digital infrastructure combining power, connectivity and land. It does not call itself an operator of data centers, because it is not one.
The gas is sized where zoning can’t reach it
The generation mix is the part that should make a project financier slow down. A Billings Gazette investigation, working from emails, put the gas-fired component at roughly 585 MW of gas-fired engines plus about 1,200 MW of gas-fired turbines, or 1,785 MW in total, alongside the renewable and storage capacity in the filing. That is a large thermal fleet for a campus whose disclosed near-term load is 500 MW.
Two readings are available and they imply different things. If the gas is grid-synchronous generation feeding the interconnection, it deepens the rate-base and cost-allocation question that follows. If it is behind-the-meter generation meant to island the campus and skip the interconnection queue for early load, it tells you the developer does not trust the wire to arrive on its own timeline, which is itself a signal about how bankable the grid path is. Either way, siting matters. The campus sits in unincorporated Yellowstone County, outside any municipal zoning authority. The mayor of the nearest town has said it is “not within our power to do anything about.” A 1,785 MW thermal build placed precisely where the closest zoning hook can’t reach it may not be an accident of geography.
Who pays for the wire, and who profits from building it
Strip the human-interest framing from the New York Times piece and the actual dispute is a cost-allocation question. The rancher at the centre of it, hedging in her own words, put a 21-mile interconnect at “US$30 million or whatever that we ratepayers get to pay for.” She does not have the FERC vocabulary. She has the invoice, and the invoice is the right unit of analysis. Network upgrades to serve a speculative, uncontracted multi-gigawatt load land on someone. The whole investment case turns on who.
NorthWestern’s Large New Load Tariff is designed to send those costs to the developer through minimum service commitments, minimum energy billing, collateral requirements and termination provisions if a data center walks early. On paper, that closes the loop. The complication is that the protections only bind a signed customer, and the entity in the queue today is a development sponsor holding an option, not a contracted load taking minimum-take obligations. A tariff that allocates costs to large-load customers does not obviously reach a 7,235 MW interconnection study that has no end customer attached to it.
And the utility is not a neutral party in this. NorthWestern’s own first-quarter 2026 materials frame data centers and FERC regional transmission as the engine behind a US$3.21 billion 2026 to 2030 capital plan and 4 to 6 per cent rate-base growth. The pending Black Hills merger would roughly double combined rate base to about US$11 billion and lift the target growth rate to 5 to 7 per cent. Transmission capex is rate base, and rate base is regulated earnings. So the entity processing a speculative multi-gigawatt request earns a return on the wire it builds to serve it. The referee in the cost-allocation fight has a position in the outcome. That is the structural fact the transparency framing misses entirely, and it is the same risk architecture I traced in the Entergy/Meta Hyperion writeup: when the utility’s growth case depends on serving the load, its incentive to scrutinise whether the load is real runs the wrong way.
The bull case, taken at full strength
The strongest version of the other side is not weak. Powered land with real interconnection position is genuinely scarce, and scarce assets command option value; warehousing queue position ahead of demand is a legitimate and common developer strategy, not a trick. EnCap is not a shell. It has US$47 billion of history, named renewable assets in the ground at the site, fibre routes toward Seattle and Salt Lake City, and a signed development agreement with the utility rather than a press release. Montana advocacy filings describe NorthWestern planning to serve three data centers scaling toward roughly 2,250 MW by 2030, so the regional demand context is not imaginary. And the tariff, if the Commission holds the line, is exactly the mechanism that would force the developer rather than the ratepayer to carry the upgrade. A bull would say this is a credible energy developer building ahead of a real demand curve, with a regulator already installing the guardrails.
Each of those is true. What the bull case underweights is sequencing. The interconnection study, the transmission planning and in some structures the upgrade spend begin years before the tenant signs, because that is how the queue works. The tariff binds at contract execution. The gap between those two moments is precisely the window in which a 760 MW utility with a growth mandate commits capital against a position that has no committed end load. The bull case assumes the guardrail and the spend are simultaneous. They are not.
Who wears the risk when the tenant doesn’t arrive
Lay the parties out. Quantica and EnCap hold the optionality: a 7,235 MW queue position and 5,000 acres of powered land that gain value if a hyperscaler attaches and can be monetised or sold if one does, with development-stage capital that is underwritten to take exactly this kind of binary. NorthWestern holds a regulated incentive to build, because the transmission and generation that serve the load convert into rate base and earnings whether or not the speculative tail of the request ever fills. The ratepayers of a 760 MW system hold the residual: if upgrades are committed and the multi-gigawatt load does not materialise, the stranded cost flows through the rate base unless the Commission affirmatively blocks it, and blocking it means overriding a tariff and a merger growth case the utility has already built its plan around.
The stress scenario is not exotic. It is simply the base rate of interconnection queues, where queued megawatts vastly exceed completed megawatts, applied to a request that is already six to seven times its own disclosed campus. If the tenant doesn’t sign, the option expires worthless for the sponsor, which is a manageable outcome for a PE fund holding a portfolio of such bets. The same non-event is not symmetric for the ratepayer who funded the early network work, or for the utility that booked the growth and now has to explain the gap.
That asymmetry is the whole piece. The sponsor is renting optionality on a tenant. The utility is renting growth on a buildout. The ratepayers of a small system are being positioned to warehouse the duration and stranded-cost risk of a load that, as of today, no one has agreed to consume. The rancher showing up alone at county meetings named the mechanism before the market got around to pricing it.
The question to take into your next diligence call on any of these powered-land plays is narrow and it is the one the Broadview filing makes unavoidable: at what point in the sequence does network-upgrade capital get committed, and is there a contracted, collateralised end load standing behind it at that exact moment, or only an option held by someone who keeps the upside and can hand back the keys?
Nothing here is investment advice.