Guide to building better AI data centers in WA

Fox - Seattle
12 followers
1 Video View·Jul 9, 2026
AI Generated·May contain errors
Making new AI data centers in a smarter and more responsible fashion. That's the aim of a new guide
for AI data center developers in a collaboration. The Pacific Northwest National Laboratory in
Richland worked with industry groups to develop the AI data center energy performance framework.
One of those groups involved was the National Electrical Manufacturers Association or NEMA.
To dive more into this guide we're joined by Senior Vice President of NEMA Patrick Hughes.
Thank you for being here Patrick. Absolutely happy to be here.
So NEMA specifically focused on electrical systems in this data center guide.
From that standpoint what are some of your recommendations to build better data centers?
Sure we want to make sure that as these facilities are being built they're being
built responsibly to be energy efficient and to be reliable and safe for the people that are
working in them and that they're contributing to grid reliability as assets not liabilities.
Well can you name some things that data center developers should be doing
right now to improve their overall electrical systems?
Sure we see these new AI data centers consuming as much electricity as some medium-sized cities
and what they can do to limit their impact on the grid and their communities is install things like
energy storage and microgrids and generation at the facility so that when they are training their
models and using lots of electricity they can actually generate that on-site and reduce their
impact on the local grid and even in times of peak grid demand think like a hot summer day
they might even be able to export power and help support the local utility.
A study from your company projects that data centers will consume quadruple the amount of
energy they do now in 2035. Can new tech and electrical systems keep up with that rapid increase?
Absolutely it can but the the pace of data center build out and the pace of innovation
has so far been faster than the ability of standards and building codes to keep up with
that and that was one of the reasons why we worked with the Pacific Northwest National Lab
and this group called ASHRAE to develop the framework because we wanted to provide that
guidance for what a responsible energy efficient safe data center looks like and that's going to be
increasingly important as the the pace of data center build continues to increase over the next
decade. Well we just keep hearing about these concerns mainly about energy consumption from
these AI data centers as we've been discussing so break down for me how can developers build
these centers without burdening local power grids and driving up the residential utility bills?
Yeah focusing on the major energy consumption for the data centers is really key and that's
the cooling piece and that's where ASHRAE brought a lot of expertise to the table around new more
modern ways to cool data centers that can consume less water and consume less energy in some cases
but recognizing that we're looking at more electricity consumption not less how do you
do that in a way that doesn't rely on your regional grid? How do you generate it on site
and limit the need for the local community to supply that data center with the electricity
it needs to operate and that's what a lot of the we call them grid interactive technologies
that we lay out in the framework are intended to help do is help those data centers be more
self-reliant so that they're not looking to the local grid to supply that power in every instance.
Well speaking of like who's helping to design what is helping to design and engineer these
ground points in your work is it AI or do you see there's going to be a bigger need for
electricians and more specifically electrical engineers in America's future workforce?
Absolutely we need more electricians we need more engineers the people who are designing these data
centers today they're doing it in a case-by-case basis in a lot of instances and it's custom work
and what that means is you're getting a range of different performance levels from data centers
from those that are highly efficient and safe and reliable to those that are less so. So when
industry gets together and you have leading scientists from organizations like PNNL talking
about the latest industry guidance and the latest building science you get a very powerful
recommendation for how to build data centers reliably and responsibly. Okay Patrick Hughes
thank you so much for your time we appreciate you coming on Washington News Wrap. Happy to do it
thank you very much.
