HELLO FUTURE: The Data Center Boom: Why the Future Needs Bigger Digital Fortresses

Speaker 1 (00:07):
Everybody's talking about data centers, so we need so many
data centers. Data centers, they're the new gateway to the future. Well,
hello future, it's me Kevin. This is a dispatch from
the digital frontier, and today we're going to explore the
front lines of the data center boom. The year is
twenty twenty five, the planet is Earth, and my name
is Kevin's really my guest today is a professor at
Virginia Tech University while leads SAD and he is the
go to guy on explaining the data center boom and
why the future needs bigger digital fortresses. Everybody's talking about
data centers. Everybody's talking about these data centers. Why they're
so important. There's a ton in Virginia. By the way,
how did you get into data center creation? And it
seems like they popped up out of nowhere all of
a sudden. Everybody's talking about data centers.

Speaker 2 (00:54):
So I come to the data centers from two angles.
One angle is I work on machine learning again, official intelligence,
and that requires data center. And in the past I
did some research more on like you know, renewable energy
and you know, management of power systems, and that also
kind of feeds into data centers.

Speaker 1 (01:11):
From my perspective, data centers kind of seem like a
no brainer to me. If we all want to have
artificial intelligence, then yeah, we should modernize our infrastructure and
we should be thinking about data centers. And if you
don't have a data center, then you can't use artificial intelligence.
And I don't want to get left behind. I mean,
America invented artificial intelligence. This is something that we should
all be celebrating as our innovation. If we don't build
the data centers, then we get at risk of falling behind.
So first things first, what is a data center?

Speaker 2 (01:45):
Data center is essentially a computing unit that allows you
to process operations at a really fast So think of
your laptop thowns thousands of thousands of faster CPUs or
what we call GPU.

Speaker 1 (01:57):
Okay, and this thing is what the size of a
the size of a warehouse. How big is a data center?

Speaker 2 (02:03):
So it can vary in sizes depending on how much
power you need, but it could be as big as
a wal Mark because you need to have multiple, for example,
processing units GPUs, you need to connect it to the internet,
so there are all sorts of add ons to a
data center.

Speaker 1 (02:18):
Most data centers though, just so people can visualize it
in their mind, they're about the size of what a warehouse?
Because I read somewhere that Mark Zuckerberg wanted one the
size of Long Island or something. He wanted to create
a huge megadata center.

Speaker 2 (02:29):
I mean it's typically a massive warehouse sized building, several
football field size, because again you need the computers, the
cooling equipment, the power infrastructure.

Speaker 1 (02:37):
Okay, so let's assume it's the basic one that's the
size of several football fields and or a warehouse. How
much can that power? Is that powering a whole city?
Is that powering just a small town? How many people
can get access to leveraging AI through a data center
of that size?

Speaker 2 (02:56):
To be precise, the data center does not power. It
actually consumes power. So the data center allows you to
run algorithms and it can do millions of operations. So
it can power Chatgypt for example. Right, and you need
multiple data centers to run the billions of prompts of
jet chatgipt that people need.

Speaker 1 (03:15):
I just had to meet the future moment, Professor Valid
Sade of Virginia Tech University, he just gave me that
meet the future moment. So at data center. It's not
powering anything In many ways, it's just a giant computer.
But if you can't plug the computer into the wall,
then you can't use the computer, which is why we're
seeing those reports in the Wall Street Journal. Folks that
in Vidio's got these data centers that are hanging out
there and they can't do anything with them because they
don't have the energy. Okay, so this is important. So
now I'm starting to understand why we're having a conversation
in this country about energy and cost of electricity. So
you build this data center, how much energy does the
average data center need?

Speaker 2 (03:59):
The average data center need the amount of energy of
tens of thousands of homes, So estimate trains between ten
thousand to fifty thousand, depending on different sources.

Speaker 1 (04:07):
See that's really helpful. This is like the first time
someone's put it that simple to me. About fifty thousand homes,
The electricity that fifty thousand homes use need to power
one data center. So the amount of electricity that Delco
uses where I grew up, Delco uses in one year
is needed to power a data center for one year.
Holy smokes. So essentially, if you build a data center
in a town. You just doubled the output of electricity
needed in that town. Something tells me that you're gonna
break the system or you're gonna have It's almost like
that cartoon where he plugs something into the wall and
then the hair just shoots up. Something tells me that
our power grid structure is not really ready for that
increase in electricity.

Speaker 2 (04:58):
It is not ready. I mean, for example, forecasts from Dominion,
which you know, powers Virginia, they say that by twenty
thirty eight, the demand from data data centers alone will
grow by like five times, right, like from two gigawat
or two point out point eight gigawa to thirteen gigawa.
So that's significant growth. And of course we can upgrade
the power system, but we need to be ready for that.
And I always like to give an analogy here with
telecom companies. Right in two thousand and seven, you didn't
have an iPhone, so you barely used your phone for
voice codes and so on, and then there was a
storm of apps within a year or two that almost
brought down our telecom infrastructure and we weren't ready. So
there's something somewhat similar here. I would say, like the
power system is a little bit more resilient than the
tel goes back in two thousand and seven, but it's
very similar.

Speaker 1 (05:44):
That's a great analogy. That's an incredible analogy. So what
are some of the lessons that we can learn from
what we did right in two thousand and seven and
maybe some learning points that we can improve upon from
two thousand and seven as we look to meet the
future with this incredible thing that is artificial intelligence.

Speaker 2 (06:05):
So I think the learning lesson is that we should
build infrastructure that is ready for things we do not foresee. Right, Like,
oftentimes we build a technology and someone asks you who
would need it? Like back to the Telco question. They
used to say, Oh, my phone works perfectly, I can
talk to my friend. Why do I need more bits
per second? Why do I need a faster teleton? Well,
all of a sudden, there was an iPhone. There was
a smart guy who created an iPhone where you can
watch Netflix on your phone. So yeah, you needed that.
So we need to be ready for something we don't
kind of envision. Second, to respond, we need to have
a research and development strategy at the national level potentially,
like what should we invest in development to short term
to meet the demand and what should we develop in
research which is potentially more important for long term similar situations.

Speaker 1 (06:49):
Professor Walid sad He is a professor at Virginia Tech University.
He got his PhD from the University of Oslo. Also
has studied at Lebanese University as well as American University
of Baber. Thank you so much for helping us understand
the data centers to meet the future. Moment that I
had is that fifty thousand homes the electricity. Fifty thousand
homes for one year is what it needs to require
a data center. All right, talk to