Oracle's Credit Rating Is Two Notches Too Low for Its $7 Billion Wisconsin Bill

The Tariff Line We Energies Wrote for Customers Too Big to Trust on Faith

We Energies, Wisconsin's largest utility, has a rule for its biggest electricity customers: any data-center developer whose S&P credit rating falls below single-A-minus has to post collateral, in cash or a letter of credit, before the utility commits to building the power for them. It's insurance against a company promising a gigawatt of demand, then walking away and leaving local ratepayers to cover the transmission and generation it ordered.

Oracle is that customer now. Its nearly one-gigawatt campus in Port Washington, part of a $15 billion investment the company has pitched as a jobs-and-tax-base win for the state, needs exactly the kind of power buildout the tariff was written to protect. Oracle's rating is BBB, two notches below the single-A-minus line, which under the tariff's math means a $7 billion letter of credit, or cash collateral of the same size, carrying a cost the Financial Times reported at more than $100 million a year.

Oracle tried to get out from under the rule in June, petitioning a county judge to strike it down. That failed, and this month the Public Service Commission of Wisconsin declined to reconsider its position too. Oracle told the FT it still expects the commission to come around once it weighs the employment and economic upside of the project, the same $15 billion figure it has used to make its case. What Oracle didn't dispute was the rating itself, and the thread noticed S&P had just cut it a notch further, from BBB to BBB-.

"One level above junk. Yikes." — a member since 2009 with 11,800+ karma

The downgrade reaches past Wisconsin. Oracle is financing its buildout on the strength of a $300 billion compute contract with OpenAI, taking on tens of billions in data-center debt against cloud revenue that hasn't shown up yet. A weaker rating means every future collateral tariff, every future lender, extracts a bigger premium for the same bet.

The Thread Doubts Anyone Is Actually Standing Up These Data Centers

The FT's numbers describe the financing side. The thread pushed on something else: whether the physical buildout behind the AI boom is going as smoothly as the pitch decks suggest. One commenter linked an industry report on delivery delays and summed up why: getting power to data centers with onsite generation is running into roadblocks "from permitting to even finding turbines to buy," concluding that "Software people are getting a lesson on the difficulties of utility scale power hardware" (an HN member since 2017 with over 9,000 karma).

A reply took aim at a fix being floated elsewhere for the same power crunch: skip the terrestrial fight and put the data centers in orbit.

"The idea that we can't get enough electrical power on earth because of supply constraints, but that we can solve that by launching the datacenters … into space makes no sense unless you're a huckster trying to pump up your space launch business ahead of an IPO." — an HN member since 2013 with 11,500+ karma

"They Laid Him and Half His Team Off"

The sharpest reversal came from someone with an actual connection to Oracle's buildout, not just an opinion on it.

"…my friend worked there in the group that went around the world standing up their new cloud datacenters, and he was there from basically 2019 to earlier this year. They laid him and half his team off earlier this year." — recounting a friend's five years on Oracle's own global data-center rollout team, before the layoff this year

The team went from roughly fifteen people to seven, not the headcount trajectory you'd expect mid-buildout on a $15 billion campus, unless the plan has quietly shifted toward contracting the physical work out instead of running it in-house.

Too Big to Fail, or Just Too Well Connected

Further down, the thread stopped debating logistics and started debating whether Oracle survives its own bet.

"Oracle is increasingly looking like the Lehman Brothers of the AI boom. One day it will all go pop and the cards will start crumbling elsewhere in the industry." — a reader further down the thread

Two replies pushed back, though not by defending Oracle's balance sheet. A reader responding directly argued Oracle is "so deeply embedded into the US government, and Larry Ellison is connected enough that they will be granted any and all bailouts necessary to prevent this," adding that everyone "pretends not to notice that such bailouts are socialism" (reply to the Lehman comparison). Another pointed to Ellison's son now running Paramount, which he's "driving it to become the next Fox News challenger," as reason enough to expect the current administration to extend the family "a sweet heart deal." A third reader waved off the collapse narrative from a different angle: Oracle could always spin off or sell the data centers to Anthropic or others at a profit, or simply rent out the compute "at insane margins."

None of those three readings needs the others to be wrong. What the Wisconsin ruling actually tests isn't Oracle's survival, it's whether other states start writing the same tariff line: collateral tied to a credit rating, not a pitch deck. If Port Washington holds, the next fight over a hyperscaler's power demand starts with an S&P report, not a ribbon-cutting.