Like you, I saw a news article last week about Oracle something something Wisconsin something something billions of dollars and moved on to the next headline.  OK, so a regulator is narrowing its eyes when it looks at Oracle because the locals are mad about datacenters jacking up their electricity rates.  Yet another story in a long line of such stories, and eventually either the people (long shot) or the lobbyists (where I’d put my money) will win.

But this story is much more significant.

Oracle is building a new datacenter in Wisconsin, and as part of that, it needs a lot of electricity.  There’s two parts to that.  First, you need the source of the juice – hydroelectric, solar, nuclear, whatever.  That challenge is significant but it’s not part of this story.  The second thing you need is all the infrastructure to haul those electrons from the Blankety-Blank River Power Authority or whatever to your datacenter.  That means transmission lines, maybe some substations, probably buying lots of right-of-ways, a lot of legal paperwork and filings, etc.  It’s not like going down to Home Depot and buying a heavy duty extension cable.

In fact, in the case of Oracle’s Wisconsin effort, it’s $7 billion dollars’ worth of stuff.

Wisconsin took a look at the project and realized that if the project fails, if Oracle pulls out, or if Oracle goes under, the ratepayers could be staring at a very large pile of unused – and unpaid for – material.  The surplus juice could be sold or retransmitted, but all the transmission equipment and setup costs aren’t going away.

Then they looked at Oracle’s credit rating.  And that’s where the fun starts.

Oracle’s credit rating was recently downgraded to one step above junk.  In a lot of scenarios, that’s inconsequential.  If Oracle owed you some money, they still owe it.  But in other scenarios, it’s hugely impactful.  And there are now 24 states – including Wisconsin – that have laws requiring financial responsibility covenants when you start building datacenters.

The way it works is that the state says “we’re not going to be on the hook for $7 billion worth of datacenter power setup, and we want you to prove that you’ll pay for it no matter what”.  If you’re Berkshire Hathaway or ADP – two huge companies with stellar credit ratings – then the state will probably take your word for it.  But if you’re below a certain rung, they won’t, by law.  They’ll require a commitment from a bank guaranteeing the project.

And banks don’t do that for free.  They charge 1% – 3% of what they’re being asked to guarantee, per year.  The riskier the project (and the company they’re being asked to guarantee), the higher the fee.  In Oracle’s case, with their near-junk credit rating, let’s assume it’s 3% (it could be more).  So that’s $210 million per year Oracle is on the hook for, probably for many years.

Now multiply that by all of Oracle’s datacenter projects.  Larry Ellison in 2024 said they’re building 100 around the world in the coming years.  Let’s assume half are in the US.  We have no way of knowing what the per-DC bank commitment fee would be, but let’s use that $210 million figure.  That’s a new $10.5 billion per year in finance expense.

Oracle didn’t have this expense until they started this datacenter buildout spree.  To give you some perspective, Oracle’s revenue is about $67bn a year.  And they have about $156bn in long-term debt.

Turns out the credit downgrade brought a lot more pain to “Big Red” than expected.  Now what happens when they need to cancel or curtail datacenter spend?  Meta has announced they’re going to start selling their vast AI compute (presumably because they bought way too much), so there’s lots of new competition.  So much of Oracle’s potential AI revenue is tied to OpenAI contracts.  If OpenAI can’t fulfill its commitments – and they’re sitting on a cool $1 trillion worth of paper promises – then we could see recursive rounds of (1) credit downgrades for Oracle, (2) costs go up to build, (3) financials suffer, back to (1) credit downgrades.

Poor Larry.

raindog308

raindog308 is a longtime community LETizen, technical writer, and self-described techno polymath. With deep roots in the *nix world, he has a passion for systems both modern and vintage, ranging from Unix, Perl, Python, and Golang to shell scripting and mainframe-era operating systems like MVS. He’s equally comfortable with relational database systems, having spent years working with Oracle, PostgreSQL, and MySQL.

As an avid user of LowEndBox providers, raindog308 runs an empire of LEBs, from tiny boxes for VPNs, to mid-sized instances for application hosting, and heavyweight servers for data storage and complex databases. He brings both technical rigor and real-world experience to every piece he writes.

Beyond the command line, raindog308 has a life-long love of German Shepherd Dogs, high-quality knives, target shooting, theology, tabletop RPGs, playing guitar, and hiking in deep, quiet forests.

His goal with every article is to help users, from beginners to seasoned sysadmins, get more value, performance, and enjoyment out of their infrastructure.

You can find him daily in the forums at LowEndTalk under the handle @raindog308.