I had an interesting chat with an East Coast utility client about what is essentially risk management when providing new service to data center facilities or office complexes that may include significant data center space.
The fundamental issue is that the utility is asked by the developer of the new facility to install the equipment that will serve the projected load, though there is often no assurance that the load will materialize. If it doesn't, there is wire in the ground, "copper on the pad" (transformer capacity), and electric capacity all the way back to the substation that doesn't generate revenue.
The results are at least two-fold. The utility now has assets on the ground that don't generate revenue, which means that rates for existing customers have to go up to cover the poor asset utilisation. The second is regulatory - the utility may get a rate base dis-allowance ruling from their regulator if these situations get out of hand.
If you want a case study of this problem, look to Pacific Gas and Electric and the dot com boom of the late nineties, when it seemed every real estate developer in the Bay Area became a data center developer.
Tens of megawatts of capacity were installed for facilities that in some cases never became data centers. (Of course many of the facilities were bought for pennies on the dollar and were eventually used many years after the bubble burst.)
Now that sounds like a cautionary tale, and it is, but there are some nuances. In California, the investor-owned utilities have a new service rule that protects them from stranded assets.
If you are developing a facility with a significant electric load, you sign an agreement with the utility to pay a "cost of service" fee if your electric billings fall below a calculated level. In other words, if you don't spin the meter enough to justify the capital outlay for facilities to serve you, you have to make up the difference.
This rule was largely put in place to cover commercial office developments. You build a high rise building on the if-come promise of finding tenants; the utility is not going to take on that risk with you.
The problem for PG&E during the dot com bust was that the companies that signed those agreements went belly up, so they couldn't collect the cost of service fees. (To this day, PG&E remains very wary of installing services for multi-MW loads, and typically negotiates very hard to install capacity in increments.)
I was surprised to find that my client on the East Coast is only now thinking of proposing a similar scheme to their regulator. In their case, they aren't seeing multi-MW stand-alone data center development, but office projects are asking for load allocations to serve potential data center space.
It would be wise for the regulator to agree to a cost of service guarantee scheme to protect the utility and its ratepayers from stranded assets.
And for developers of utility-scale data centers, you may find it easier to work with your utility on a phased approach to securing electric service capacity: build your substation yard with switchgear, then install transformer capacity in five or ten megawatt increments as you build out your site. Photos of Apple's new data center in Prineville suggest they are doing just that.
Comments