Power Grab: Data Centers and the Grid Part 4 - AI & Wholesale Power Markets

What if the explosive demand for data centers could transform power grids across the nation? Join us as we unravel the latest developments in data center supply strategies and the energy market's evolving dynamics. From Vista’s power sales discussions with major companies to California's innovative district energy system marrying data centers with residential units, this episode is packed with groundbreaking insights. We explore the potential of modular nuclear startup Oklo’s partnerships and the intriguing collaboration between Sharon AI and New Era Helium Corporation for a new data center venture in New Mexico. Plus, Bloom Energy’s significant deal with AAP Ohio takes center stage as we discuss its implications.

As we navigate the intricate web of these developments, we turn our attention to their broader impact on electricity users and wholesale markets. The rapid growth in data center demand is reshaping the power grid landscape, particularly in the Mid-Atlantic and PJM regions. We delve into the potential consequences for bulk power grids, especially with hints that advancements in large language models may be slowing. Could this mean an overinvestment in capacity infrastructure? We tackle these provocative questions and more, offering a critical perspective on the challenges and opportunities in the ever-evolving energy and data center sectors. Tune in to stay ahead of the trends and anticipate the changes on the horizon.

Support the show

Episode Transcript

Speaker 1 (00:00):
Last week we talked about data center supply
strategies.
Well, a few updates came to myattention in the ensuing week
since we've spoken last.
The space isn't slowing downanytime soon, it seems.
First, vista announced it's indiscussions with two large
companies concerning sale ofpower from existing gas and
nuclear facilities, with plantsin ERCOT and PJM eliciting

(00:21):
interest.
One option may be co-locationdeals at multiple sites that
might involve construction ofnew assets as well.
Second, in a unique project,california-specific gas and
electric and real estatedeveloper West Bank planned to
develop a district energy systemcombining three new data
centers using up to 200megawatts and as many as 4,000
residential units in San Jose.

(00:42):
Excess heat from the datacenters will yield heating and
hot water to neighboringbuildings.
Further details have not yetbeen shared.
Third, modular nuke startupOklo says it has letters of
intent to partner with two majordata center providers and
deliver up to 750 megawatts ofpower.
Fourth, sharon AI andindustrial gas company New Era

(01:04):
Helium Corporation haveannounced a non-binding letter
of intent to form a JV for thedesign, development and
operation of a 90 megawatt datacenter at a site in New Mexico's
Permian Basin.
New Era already produces heliumnatural gas and dry natural gas
and plans on supplying some ofthat gas for an on-site
generator that would support theplant.

(01:26):
And fifth, last week we lookedat the option of fuel cells and
I commented in the face of tensof gigawatts being sought, these
quantities are akin to a pimpleon a fruit fly in terms of
significance to anybody, butBloom Energy.
Well, bloom did score big inthe intervening week, announcing
a deal to sell up to onegigawatt of fuel cells to AAP

(01:47):
Ohio.
Still small in the scheme ofthings, but pretty significant
for Bloom.
Okay, back to business.
We've looked at the deluge ofdata center interconnection
requests totaling well over100,000 megawatts and rising.
We've looked at why this ishappening, the AI arms race, and
we've touched on possiblesupply strategies.
Now let's discuss the potentialimplications for electricity

(02:08):
users.
In this session we'll talkwholesale markets and next week
we'll cover distributionutilities.
But first one other quickupdate from last week.
I mentioned that gas pipelineenergy transfer was having
discussions with data centersrepresenting 3 billion cubic
feet of usage per day.
Well, it's just releasedNovember 12th numbers indicate

(02:29):
it now has requests to connectto over 40 data centers in 10
states, total gas consumption Upto 10 BCF per day.
To put that in perspective,last year's national usage was
89 BCF per day, and one morecontextual number 2023 US LNG
exports total 11.4 BCF per day.
So this is big stuff just fromone gas pipeline company.

(02:53):
And one more thing just to castsome high-level doubt into the
equation Very recent commentsand indicators from the industry
hint that just maybe the gainsfrom these large language
learning models are starting toslow down.
Indeed, one trade journalreported last week that GPT
improvements are slowing down.
This may be a result of ashortage of data for models, but

(03:16):
the implications could be huge,because they imply that simply
throwing stronger chips and moreenergy at the problem may not
be enough to scale these modelscommensurately.
We may end up building lots ofcapacity infrastructure with
limited economic returns on themargin.
The terms-granted assets shouldcome to mind, but nobody knows
just yet how the technology isgoing to play out.

(03:38):
But let's ignore that for nowand assume the tequila bottles
will keep coming and the AIlanguage learning model training
party is going to go on fullblast.
If that's the case, what wouldbe the likely impact to our bulk
power grids and wholesalemarkets?
The answer to that is nuanced,in that some markets will be
much more effective than othersthe northeastern power markets,

(04:01):
iso, new England, new York ISOthey likely won't see that much
of an effect.
Land is limited there and theenvironmental regs are fairly
tight, making it harder to siteeither data centers or new power
plants to supply them.
The mid-Atlantic states and PGMare another story.
In January PGM had tripled itsforecasted growth numbers from
last year.

(04:22):
That was well prior to a lot ofnew utility announcements, so
the next update in two monthswill likely get a lot higher.
We all know of the pressure inDominion Service Territory, the
global data epicenter, but thatload seems to be metastasizing
to other regions.
New Jersey's PSEG, for example,just filed an updated forecast

(04:42):
to PGM in late October,projecting data center load to
grow from today's 343 megawattsto 1196 by 2030.
And Exelon, in late October,requested of PGM a large load
forecast adjustment of 2600megawatts through 2029.
The mid-continental ISO, miso,is not seeing quite the same

(05:04):
pressure, but its CEO noted thismonth that 2,500 megawatts of
new data central load showed upand over 4,600 megawatts of data
central load is noted in MISO'sJuly 30th existing large load
and new load additions update.
So there's some pressure theretoo, but not compared to Texas,

(05:25):
which sits in a league of itsown, with a potential near
doubling of existing demand.
Encore alone, as we notedpreviously, reports 59,000
megawatts of requests.
In ERCOT's system.
That peaked at 85,000 megawattsthis past summer.
And then what about SPP, theSouthwest Power Pool in
California?
Well, spp is not a big datacenter market yet, but it is
growing.
Politicians in North Dakota,for example, noted that they may

(05:48):
see massive growth from twoundisclosed data center
companies, with growth startingat 500 megawatts, going to a
thousand megawatts and perhapsgrowing 10x from there,
involving up to $250 billion ofcapital.
For its part, and despiteSilicon Valley being nestled
there, california's power pricesmay be just too high and land
too expensive to attract much AIload.

(06:09):
So, from a heat map perspective, the pressure is most likely
going to be centered oncompetitive markets in Texas and
PJM.
But power prices will beaffected everywhere for one
principal reason All this newload and the supply to meet it
will be facing a shortfall ofequipment such as transformers
and switchgear and the rawmaterials that go into them

(06:30):
Copper, specializedgrain-oriented electric steel
for the transformers, thosesorts of things.
Add to that likely futuretariffs on imports from China,
from which we source billions ofdollars of transformers, and
demand has the great potentialto outstrip supply.
There's already a three-yearwait or longer for some types of
transformers.

(06:50):
Add it all together and youhave a general recipe for an
inflationary energy environmentacross the entire country's grid
, irrespective of where thosedata centers are going to be
located.
Then pile on top of that thefact that interconnection queues
are stodgy and you have a veryinelastic supply function.
In other words, it doesn'tmatter what the buyer is willing
to pay, the supplier is simplyunavailable.

(07:12):
Lawrence Berkeley Laboratory'slatest interconnection report
indicates that the medianduration from transmission
interconnection requests todelivery at first megawatt hour
is now at five years.
So where are prices headed?
Demand is exploding and supplyis scarce and more expensive.
Power prices are going up morein some markets than others.

(07:34):
Consulting group Bain Companyprojects average annual cost
increases of 1% annually.
My conjecture is that theseincreases will be concentrated
more in some areas of rapidgrowth than others where demand
is more tepid.
Some competitive power gridswill see more marked price
increases than others simplybecause of the ratcheting up of

(07:54):
new demand, but nowhere willthis dynamic be deflationary.
In the next and final sessionwe'll talk about the
distribution utilities, theenormous uncertainties facing
them in this build-out and theinherent risk of overbuilding in
an extremely uncertain AI world.
Well, thanks for watching andwe'll see you again next week.