How "Materially Smaller" Changes the Numbers...

For anyone who's been paying attention, it's obvious that the City has substantially increased public expenditures in its hell-for-leather pursuit of this data center project.

The City has borrowed money at an alarming clip, racked up significant legal and PR expenses, given away public land it valued at $17.5 million for $200, agreed to significant tax incentives and other infrastructure commitments, and committed money to other projects that I'm sure they're banking on future growth to help pay for down the road - i.e., the Marc Mahal.

The problem is that the version of the project the City spent all that money pursuing isn't the same project that's now under consideration.

The scale of the original proposed Matrix Project was ridiculously enormous. The illegally-passed agreements contemplated three phases totaling 3,000 megawatts of capacity, required the developer to add improvements with a minimum taxable value of $4.9 billion by September 2027, and ultimately required $18.7 billion in minimum taxable value before the agreements expired.

Now, I never believed a 3,000 MW data center with $18.7 billion in taxable value was remotely realistic, and certainly not with MSB Global Fraud as the developer. But what I believe isn't really the point.

The important thing is that the City apparently believed it (unfortunately for us).

That's the project they were basing their economic expectations on. That's the scale they were using when they were talking about all of this transformative revenue, and that's the project against which they justified the land, infrastructure, debt, incentives, and all the other costs that have accumulated around this thing.

I don't give much weight to salesman math from anyone, let alone Marc Maxwell, who at various points was throwing around obscene revenue projections. But we don't need to accept his math to recognize the obvious: hypothetically, an $18.7 billion taxable project would generate a tremendous amount of revenue.

At that scale, you can absorb a lot of bad decisions. However, you can't turn a profit on crazy ideas alone.

The new potential developer, CyrusOne, has publicly said that it is considering a project materially smaller than the Matrix concept. That's a very important change because “materially smaller” could mean almost anything. Are we talking about 500 MW or something even smaller? More importantly, what taxable value gets put on whatever CyrusOne ultimately proposes? Because it isn't $18 billion.

This matters enormously because once the project gets materially smaller, the economics of the deal change drastically as well. You don't get to shrink the taxable value and pretend the economics stayed the same.

The land doesn't come back. The roads don't unbuild themselves. The water and sewer lines don't disappear. The bond interest doesn't stop accruing. The legal bills don't get refunded. The years already lost don't get added back onto the agreement.

The fundamental problem is this: The City has spent money like it was getting an $18.7 billion taxable project, and now it clearly is not. So what does that do to the financial components of the deal?

When you start adding it up, you see the scale of the public commitment involved here: $17.5 million for the land, about $13 million in infrastructure and public capital, and another say $3 million in estimated legal, lobbying, PR, consulting, and administrative costs.

That produces an economic investment figure of about $33.5 million.

The land figure is also an economic value rather than a cash expenditure. The City owned an asset and committed it to this project, and giving away an asset has an opportunity cost just as spending cash does.

Then there is the debt.

In 2024, the City authorized roughly $12 million in certificates of obligation for Thermo-area streets, drainage, water and sewer infrastructure, and related professional costs. That means the financing itself adds roughly another $10 million in interest costs if the debt is carried through maturity.

Depending on how you look at it, then, the City has committed roughly $33.5 million in economic resources and could ultimately incur roughly another $10 million in financing costs over time. And the lower the taxable value is, the harder these numbers hit.

Another problem is that the revenue is nowhere in sight.

That timing of the revenue matters a lot because the existing agreement doesn't run forever. The current deal expires in 2037, so every year the project gets pushed back is one less year in which the City has the benefit of the investment requirements it originally negotiated.

So, the debt obligations started in 2024. The revenue may not really start until 2030 or 2031. Meanwhile, the current agreements expire in 2037.

That's a very unfavorable mismatch.

So what happens after 2037? We don't know. Nothing in the current agreement requires the developer to keep making investments after the agreement expires, or to maintain any particular level of taxable value beyond that point.

CyrusOne may continue reinvesting because it makes business sense to do so, but that decision would be theirs. The City would be relying on continued private investment that it has not contractually secured, while its own debt obligations continue for years afterward.

In other words, the optionality belongs to the developer, while the long-term obligations belong to the City. I have seen nothing that guarantees that CyrusOne will maintain a particular taxable value after 2037.

Another issue is what kind of taxable property a data center actually creates. A data center has real property - land, buildings, and permanent improvements - but it also has enormous amounts of business personal property, including servers, computing equipment, and networking hardware, which is taxable to a degree.

That distinction matters here because TIRZ #2 captures 75% of the City's tax increment attributable to real property, while business personal property is treated differently.

So, how can we get an idea of what the CyrusOne "materially smaller" data center might mean economically? We can make some back-of-the-napkin assumptions like:

• 35% property-tax abatement for the first 10 years of operation

• 37% real property and 63% business personal property split in tax (based on what I've found other places, but it's TBD how Hopkins CAD will do this)

• First full taxable year beginning in 2030 or after (the Vistra litigation isn't likely to be resolved any time soon)

• A modest 8% rate of return (it should be much higher considering other risks to the environment, infrastructure, natural resources, etc).

We can then plug that info, along with the other things we know, into a model and see what taxable value is required when adjusted for the variability of post-2037 reinvestment and revenue timing. It looks something like this:

What you can see here is that the taxable value, reinvestment assumptions, and revenue start date all have significant impacts on whichever financial model the City is using to justify its economic assumptions.

A $2.3 billion project can look at least financially acceptable if you assume the tax base remains substantially intact for decades. The same project looks terrible if you don't give the City credit for future reinvestment that CyrusOne has never promised to make.

And delay makes it worse, which makes the idea of committing the resources before the legal challenges were resolved seem incredibly costly and foolish.

In the worst case, where I give the City no credit for post-2037 value, a project reaching full taxable value in 2030 needs to be around $4.8 billion to earn the modeled 8% return. If full value slips to 2031, that rises to roughly $5.6 billion. Push it to 2032 and the required value is about $6.7 billion. Higher RoR requirements push it up. Fewer incentives to the developer push it down.

The problem is that we don't know which of those scenarios we're talking about. And that's where I keep coming back to transparency. The City should have its own spreadsheets.

Before giving away land, borrowing money, building infrastructure and approving incentives, the City should have modeled projected taxable value by year, real property versus business personal property, depreciation, expected equipment replacement, tax abatements, TIRZ restrictions, sales and use tax, utility revenues and costs, debt service, public-service costs and alternative uses of the land and money. In fact, the City's own policies contemplate much of this information.

So where is that analysis?

Also, the point of this is not for me to try to tell you that there's a number where I think a data center here is viable. It's not, because viability is about more than money. I have absolutely no confidence in this City Manager and this Council to negotiate a good deal on behalf of its citizens and protect them from the downside risks of a project like this.

There may be some other context in some other place where a data center might make sense. But it's not here. Not with incompetent people at the helm. Not within a community that clearly doesn't want it and is fed up with the City's reckless pursuit of it.

Still, maybe CyrusOne ultimately proposes something large enough to justify everything Sulphur Springs has already put into Thermo.

If so, show us the numbers. “Data centers are worth billions” isn't an economic-development analysis, and “think of all the tax revenue” isn't an investment model.

The questions now are simple:

• How much taxable value will CyrusOne actually put on the tax rolls?

• How quickly will it get there?

• What incentives will the City give away to get it?

• How long can taxpayers reasonably expect that value to remain?

Until the City answers those questions, nobody can credibly say the revised project is a good investment.

They can only say they hope it is. And, as we've seen, hope is a particularly expensive strategy.