Build1 distinct publisher3 min readUpdated
We Energies' very large customer class makes any sub-A- offtaker post guarantees. S&P's July 9 cut to BBB- put Oracle three notches under that line.
The Engineer · Build desk

Compiled by The EngineerSomething wrong?How this is made
On July 9, S&P cut Oracle's issuer credit rating from BBB to BBB-, one step above junk [1]. The next day, according to an account published on dev.to, Wisconsin's Public Service Commission let a deadline expire without putting on its agenda a request to revisit the rule that prices exactly that downgrade: under the "very large customer" structure the commission approved in April for We Energies, any customer in the class rated below A- must post financial guarantees before the utility will sell it electricity [2][3].
For the Oracle subsidiary co-developing the Port Washington campus, roughly a gigawatt on 672 acres alongside Vantage and OpenAI, the same account puts the bill at more than $100 million a year in cash deposits or letters of credit before the project can proceed [4][5]. Spread across a gigawatt of contracted capacity, that is upward of $100 per kW per year [6], paid not for energy but for permission to buy it.
The litigation predates the downgrade. Oracle sued the commission in Ozaukee County Circuit Court on June 19, arguing the PSC acted outside its authority and on insufficient evidence, and that the A- line is not needed to prevent harm to anyone [7]. When it filed, Oracle sat two notches under the threshold; three weeks later it was three [8].
The utility is on the customer's side of this. We Energies asked its own regulator on June 10 to back off the rule, warning it would push investment out of the state [9], and its lawyers argued that "tens of billions of dollars in Oracle's value would need to be destroyed before creditors or counterparties, such as Wisconsin Electric and its other customers, could experience losses" [10].
That is a statement about a balance sheet, and the commission is not underwriting a balance sheet. We Energies now bills large data centers directly for the generation built to serve them, and that generation includes the proposed Red Oak Ridge plant in the town of Paris, a build costing north of a billion dollars [11][12]. A gas plant is a thirty-to-forty-year asset; the demand it is being built against is about three years old, a ratio of roughly ten to one on asset life over demand history [13]. If the tenant walks, the plant stays, and the residual lands on schools, small manufacturers and residential meters in southeastern Wisconsin [14].
The credit file moved in the same direction while the reconsideration request sat. Oracle's AI borrowing had pushed debt-to-equity past 400% as of May [15], and the stock shed more than $50 in the month before the deadline [16]. S&P's note named OpenAI as a key credit risk inside Oracle's $638 billion backlog [17], projected a free operating cash flow deficit of roughly $42 billion for fiscal 2027 [18], and cited about $167 billion of debt already on the books [19].
Wisconsin is not an outlier. As of May, twenty-three states had approved at least one large-load tariff and another seven were pending [20], thirty jurisdictions in play [21]. Virginia's version of Dominion's GS-5 tariff begins applying automatically to customers at 25 MW and above in January 2027, with minimum fourteen-year terms [22].
Watch the Ozaukee County case for whether a court will treat a rating threshold as within a commission's authority [7]. Watch whether the tariffs already approved in other states import credit floors rather than just demand charges and term minimums [20][22]. And watch how quickly developers restructure offtake so the signing entity is not the one carrying the downgrade.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
The Oracle subsidiary co-developing Port Washington is facing more than $100 million a year in cash deposits or letters of credit before it can move forward.
On Thursday, July 9, S&P cut Oracle's issuer credit rating from BBB to BBB-, one step above junk.
On Friday, July 10, Wisconsin's Public Service Commission let a deadline expire without putting on its agenda the request to revisit a rule it approved in April as part of We Energies' new 'very large customer' rate structure.
Under the rule, any customer in We Energies' 'very large customer' class rated below A- must post financial guarantees before the utility will sell it electricity.
The Port Washington campus is roughly a gigawatt on 672 acres, co-developed by an Oracle subsidiary alongside Vantage and OpenAI.
Oracle sued the commission in Ozaukee County Circuit Court on June 19, arguing the PSC acted outside its authority, on insufficient evidence, and that the A- line is not needed to prevent harm to anyone.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Specific and dated, but single-publisher and largely unattributed
One dev.to post carries the entire cluster. It offers unusually concrete, checkable specifics - a dated rating action with three S&P figures, a dated utility reconsideration request with quoted counsel, a dated court filing and venue, tariff thresholds and effective dates - which lifts it well above rumor. But there is no corroborating publisher, no linked docket, tariff sheet, or S&P note, and the load-bearing commercial number (more than $100 million a year of collateral) is asserted without an attributed calculation. The analytic core about duration mismatch and stranded cost is the author's inference about regulator motive, not quoted reasoning.
Mechanism in force and spreading, outcome unresolved
The credit gate is not a proposal: the rule was approved in April, the reconsideration deadline lapsed on July 10 leaving it in effect, We Energies already bills large data centers directly for dedicated generation, and comparable regimes are widespread - twenty-three states approved with seven pending, and Dominion's GS-5 auto-applying at 25 MW from January 2027. What is not yet adopted is any resolution at Port Washington: the campus is blocked, the litigation is pending, and no source shows collateral actually posted or the project proceeding.
Framing runs slightly ahead of what is shown
The verified factual core - a downgrade, a lapsed deadline, a live sub-A- collateral clause, pending litigation - is reported soberly and is arguably under-covered elsewhere. The overstatement is in framing: the headline thesis that a letter grade is the single binding constraint rests on the author's assertion that everything else 'lines up,' with megawatts, queues, lead times, and water dismissed rather than evidenced, and on an unattributed $100 million-a-year figure amplified rhetorically. The regulator's motive (pricing duration mismatch) is imputed, and the Bonneville smelter analogy carries more argumentative weight than the Wisconsin record supplied.
Both quoted parties are self-interested advocates
The record supplied is built largely from adversarial advocacy. Oracle is a litigant seeking to void the threshold, and We Energies - which wants the load and bills the customer directly for over a billion dollars of new generation - petitioned its own regulator to drop the rule and supplied the quoted credit argument. Those incentives are visible on the face of the source. The publisher side is a personal-platform post that cross-promotes the author's own earlier writing on FERC and scarcity; no financial interest is disclosed either way, so publisher motive is inferred only from that self-reference, not asserted.
Plausible and specific, but uncorroborated
Confidence is capped by structure: one publisher, one article, no primary documents attached, and several load-bearing elements (the collateral magnitude, the regulator's rationale, the stranded-cost outcome) resting on authorial inference. Internal consistency is good and the dated hard facts are the kind that would be easy to falsify, which supports a middling rather than low score, but nothing in the cluster independently confirms any of it.
product
Nvidia's $105bn guarantee, not OpenAI's balance sheet, is what makes Ohio's 8 gigawatts buildable1 distinct publisher
product
Siting Is Now a Schedule Risk: 75 Data Centre Projects, $130bn, One Quarter1 distinct publisher
invest
The rolling bubble: stop asking whether AI is one, ask which layer is repricing2 distinct publishers
product
Data centers are now a ballot issue in 40% of US races, and roadmaps have not priced that in1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · August 18, 2026