Welcome to Let's Get Digital.
I'm Keri Charles, your host, and so glad that you're here with us today.
I have a very, very special guest.
You're going to love this episode.
So get a cup of coffee, sit back, and get ready to be inspired.
I have with me today Chris Sharp.
He is the chief technology officer for digital realty.
Chris, thank you for joining me.
Thanks for having me, appreciate it.
We're looking forward to it.
Yes, yes.
So Chris, I always like to ask the guests about their journey.
How did you get into this industry?
Maybe a couple, maybe a cool story.
Yeah, no, I mean, I'm a technologist at heart, right?
I absolutely love technology, you can see behind me.
um It's what drove me, you know, not only to this industry in particular with digital
reality, but just overall infrastructure, right?
So early career, starting at a network, evolving to security, and then seeing the true
frontier of data centers before they were cool.
So absolutely.
Oh, that's awesome.
Yeah, and I know that you've been with the Digital Realty now.
How long again?
10 years, coming on 10 years.
So, I'm not.
digital reality to us.
I'm sure the entire world knows who you are, but...
Yeah, no, I appreciate it.
Like I guess it's a couple of perspectives, right?
And I'll kind of go a little bit, maybe deeper in, you know, helping, you know, form some
of the largest networks in the world where Quest.
So I started at SP Southern Pacific railroad telecom building up Quest and then Quest, you
know, it was pretty interesting in what we were able to do with all of the optical
interconnection building these massive, you know, networks before.
people even realized how they're going to be utilized and then evolving into a couple of
other companies, one of which is bringing MCI out of bankruptcy and making sure they
understood how that was going to be utilized and those assets.
you know, what I saw early on is that networks are important, but how do you get more
relevance around the networking and internetworking environment?
And so I built the managed security practice at MCI.
And so I ran that and then sold MCI to ultimately Verizon and
you just see these technological evolutions.
And so in running that managed security practice, you really saw the inner workings of
broader value and interdependency on the network.
And so then where I saw that evolving is that the networks were being a bit
disintermediated from the value of what the customers were trying to achieve.
And that's what brought me to, before digital, Equinix.
And so we saw the value and the step functions of networking, internetworking,
And quite frankly, you see with data centers, a lot of the value is created with
interconnectivity.
And so I wrote a thesis when I left Equinix almost 10 years ago in that you want a broader
wallet share, right?
Comprehensive product portfolio, which you see it being written about all over in the
recent news.
um You need to be able to solve a broader capacity block with power density.
And so I wrote a thesis on, hey, a broader full scale capability with digital reality.
being able to provide wholesale and co-location.
So cab, cage, hundreds of megawatts, all heavily interconnected.
So the networking still fits within that, Kerry, and that's some of the stuff that we see
these stepping stones of my career in how technology has driven value in solving more
complex problems for our customers.
And that's kind of the stair-step crescendo that I got to digital reality.
And what makes digital reality a bit differently is not only
the full scale product portfolio, but really what we're seeing in a global platform,
right?
Where most companies come to us today, they want to design once, deploy many.
So we have the ability to not only solve for larger problems around power blocks, but
density.
And we can talk about AI and go into detail where density of the power in the data center
is a big differentiator for us at Digital, really.
But then how do we have an open platform to interconnect
all these critical destinations, be it cloud, be it AI, be it at the epicenter of solving
for some of our largest customer problems around, you know, how can I get faster
interconnects of power and how can you have long-term durability of growth with working
with utility operators and then the enterprise.
They don't want to buy a science project.
They want to buy it out.
So private AI is the other element that truly differentiates us at digital reality.
Interesting, private AI.
um I haven't heard that before.
I have heard a lot about inference, and I wanted to ask you, what is the difference in
training versus inference workloads?
And also, where does digital reality focus?
Yeah, I love it.
so many of our listeners today have probably like heard a ton about like training and
inference, but just to make sure everybody's with us is that training is the initial setup
where you have a monumental amount of infrastructure trying to pull data sets together to
train these frontier models.
There's few individuals doing that.
There's a smaller user group that are absolutely dead set in doing that.
We don't invest our capital there.
We're really watching how AI is going to be consumed.
And that's where in the most simplistic sense, it's called inference.
so watching how inference is brought to market, that's where we come into play.
And inference, what we're seeing as the complexity and the maturity of these models evolve
with like reasoning and some of these other capabilities, it has proximity to the center
center, city center.
And so what we're doing is we apply our capital closer and closer to that city center.
And that's where we're at the epicenter of allowing that inference to be utilized in
interconnected to a broader AI ecosystem.
And so it's not only about that inference.
I'll go one step further.
How do you monetize it?
Because I think that's where a lot of people are starting to think about how is AI going
to truly be monetized and consumed?
And that's where we work and understand the specific workloads for our customers and how
they can get access to the infrastructure and consume that.
But that's where this crescendo of infrastructure, where as much as
you know, we bring to market in a capacity model of training, it'll be 10, 10 X with what
we're going to need for inference.
And so you have to really watch what is the durable workload that's going to be deployed
in the environment and how is that going to be interconnected?
And so that's where we're really watching and looking at the market.
And just for the highest reference point where we've been watching is generative AI, the
first true
monetization to the enterprise is copilot with Microsoft.
And so you can follow how that's been evolving for some time.
You've been in the industry quite a while, so I really would like to get your perspective
on the evolution of the industry.
uh first, my question is, what is the current and future state of colocation?
Absolutely.
um Colo is changing drastically, right?
Where, you know, I always joke a little bit and I'll take a step back if it's all right,
Kerry, that if you learn a lot from what we did with cloud, right?
Like, so one of the things that really, you know, was a formidable step in my career early
on at Equinix was, hey, is cloud friend or foe?
Right?
Like we were all being asked that.
And so we were a...
co-location only provider and so many of our investors were like, my gosh, what is going
to happen with the company?
And so I was, I had the opportunity to build a team called the cloud acceleration team.
And jokingly, we called it the cat team, but that cat team, what we did is understood the
fundamentals of how that technological evolution was going to impact the market.
And what we were able to do is private consumption of cloud.
Like don't overthink it.
That was hybrid, right?
Where you have
private environment to a public cloud and removing the complexity from that
interconnectivity.
We built the cloud exchange.
That was a great formidable model, right?
That changed the way we viewed co-location at that time to the core of your question.
AI is very similar.
In AI, this wave is bigger and faster than anything we've ever seen.
Like the growth trajectory, if you look at what we saw with cloud, which was tremendous
growth in the industry, it's nothing compared to what we see with AI.
And so with the way that artificial intelligence is changing co-location, you need to
think about not only large capacity blocks, but the densification.
So what we're seeing in the changing of co-location is that you have to be able to ideate
with your customers.
And I referenced earlier, Kerry, private AI is where lot of our customers, our enterprise
customers have aspirations, is that all of artificial intelligence starts with your data.
And so I'd invite everybody to go read about data gravity.
We have some papers on our website, very simple, market texture kind of concept where you
want to really understand where your data is being generated and then where you can store
your data efficiently and then you want to act off that data.
so colocation, it's all about data and now with AI, it's about bringing algorithms to that
data.
And so that's where we're at the epicenter now is can you support and with this
accelerated compute, be it Nvidia, we have great partnership with Nvidia.
Can you support that hotspot and bring in that accelerated DGX pod like we've done with
Gephian and a lot of our other customers, bringing that infrastructure closer to their
data where you can action off of that quicker and it's not a science project because it
gets very complicated very quickly.
So you as a co-location provider and we, the amount of heritage we have in engineering at
this company is phenomenal in that.
you really are changing the landscape of how colocation is viewed where larger capacity
blocks being able to densify and bring that to existing infrastructure where they can act
off of not only their own data, but the data oceans that are represented with the public
clouds.
That's what's changing colocation today is you need to not only look at the chips that are
being delivered today, like one piece that you may like to know, Carrie, is like from
Hopper to Blackwell, that one simple step function with which is roughly 12, 12 months.
it's 30 times more efficient in doing inference, 30 times.
But it requires liquid, it requires a different type of design.
So co-location providers such as ourselves, you have to bring liquid to the floor.
We're highly, highly built for that.
Our asset class coming from scale allows us to do that modularly.
But that's what's changing the landscape for co-location today, Kerry.
Why are not all data centers created equal?
love it.
Love it.
Well, it depends on your heritage, right?
Like I spoke earlier, we spent a lot of time like early in my career and a lot of people
laugh at this if they know me.
Like when I left Equinix and I came over here 10 years ago, people are like, why is a
technologist working at a real estate company?
mean, our largest investor first week stepped me down and she's awesome.
I absolutely love her.
She was like, what are you doing with your career?
And I'm like, you know, like, like bear with me.
Watch it will change over time.
And we were joking early days on how that was going to come to market in that, you know,
she was also the one that, you know, referenced that in 2017, I put a slide up, Hey, AI
and digital really are going to be the home of AI, right?
Because of the capacity blocks, the master planning that we do, and because of the
modularity of our designs, which is the core of your question, Carrie, that's what
differentiates data center operators.
And that if you anticipated
the fact that the infrastructure is going to ideate rather quickly, requiring a different
type of electrical distribution, requiring a different type of cooling, and then
ultimately requiring a different type of interconnectivity in an open platform.
Those three elements truly differentiate different data center operators, where you can't
bring that electrical distribution expertise, you can't bring that type of cooling, where
it then really stifles the customer's ability
quite frankly, get the benefit out of some of the new technology coming to market.
And that's why I reference the step functioning capabilities.
That's what's really coming down to what really differentiates certain colo environments
from others.
What does the term, Braggowats mean?
I remember, dude, said that once and I thought, what is that?
Yeah.
So like Andy and I joke, our CEO.
yeah.
So there's a lot of noise, right?
Karen.
I think that's probably why a lot of people are listening to us today is like, what's
going on in the landscape.
my gosh.
It's just, you know, with AI and we have access to information and information's wildly
available, but not all information is correct.
There's a lot of hallucinations in the information, but what we, what we really talking
about with Braggawatts is you read every day.
Here's a gigawatt there.
There's a gigawatt there.
Here's a hundred megawatts there.
The fundamentals of what it takes to bring this infrastructure to market is not easy,
right?
The entitlements, the utility interconnects, the substations, like we operate in pretty
much every major Metro on earth.
And so that heritage and expertise allows us to really execute on those programs.
What Braggowats is about, don't leave everything you read.
There's going to be a lot of, you know, fully, you you read the press release and my gosh,
there's going to be a gigawatt delivered in this market.
it's not gonna come to fruition.
And that's where we've worked and we have the privilege of supporting some of the largest
customers on earth.
um And so really showing them not only what we deliver for them today, but that runway
where they want three, four, five X and growing in that market, that's what we're really
focused on.
And so the Braggowats is just don't read everything you, or don't believe everything you
read.
I think that's a good lesson for all of life, right?
So uh let's talk about the question that's uh really on everyone's mind in the data center
world.
uh With data centers already straining grid capacity due to high power demands, and also
what you just said with AI, really accelerating that demand even more.
How can the industry adapt?
And are there practical solutions?
that can really balance that growth with the grid resilience.
100%.
Yeah, power is a critical ingredient for our collective success.
And so it takes a village, right?
Like I speak and I get the opportunity to spend a lot of time with our utility partners
where it's just for the most simplistic sake.
It's not only generation, it's distribution, right?
We're the grid.
And we're always ensuring that we're collectively thinking about the problem.
And because like when I reference Braga Watts, there's a lot of demand that quite frankly,
the utility operators are like, that doesn't make sense, right?
Where they don't believe that the actual utilization will be near the levels that people
are requesting.
And so that can create a huge, you know, bubble in the industry, if you will, that if
somebody comes in and they say they're going to take down a certain amount of power, the
utility builds it.
That's a bad outcome if it doesn't get utilized.
And what really we look at is that I referenced it earlier.
long-term durable workloads that we have, you know, 10 plus year contracts with our
customers today, they really, and they're some of the top credit worthy customers on
earth, we represent a different type of demand model to the utility where they get
comfortable with that.
And so that's where it takes a uh understanding of the existing kind of landscape of
utility operators, but then you have to start thinking about Horizon too.
And so Horizon 2 is where we start to work with utility operators and some new entrants
into the market around nuclear.
I often get asked, and I've been talking about this for some time now, many years, and
that a reporter, think it was from the BBC, is like, what do I not know about data
centers?
What's going to blow my mind?
I'm like, well, we're going to have to use nuclear to meet the demand requirements of the
power associated with that.
And so there's traditional types of nuclear in large format.
There's small modular reactors in smaller format.
I'm pretty excited about SMRs.
You can tuck them in the grid to overcome the distribution challenges where right now
there's a lot of generation, but it's an islands.
It's very hard to get to certain locations because we don't support training.
We can't move the workload out there.
You really want to have the actual workload closer to the city center, closer to that
ecosystem so that that infrastructure gets more value.
So we're always looking at and working with the utility operators and
You know, we build in some of the most storied markets on earth like Ashburn, Virginia,
where we work with our partners at Dominion.
So just watching how we can not only meet that demand, but they'd be intelligent.
And I think that's a core part of what you're asking, Kerry, is we have to be more
transparent in what the workload heuristics are going to do.
Like, how is it going to operate over some period of time, really signaling to the
utility operators in building a partnership where it's not two silos that are not directly
interconnected and directly communicating to one another so that we can like collectively
solve the challenges of some of the peak loads that are existing within this grid.
And my last piece is just being a good partner, right?
Not only being a good partner to the utility operator, but the Metro, right?
So when we operate in a market and a lot of the markets we're investigating peak shaving.
So as you can imagine with anything, right, you have to build for the peak.
And so with some of the utilization, with some of the heat waves we see today, one of the
things we do, and it's a little misunderstood fact that with the roughly three gigawatts
we operate today in IT load, there's three gigawatts of diesel generation behind that.
And so we can do a lot of peak shaving where in critical moments when the demand spikes,
we can shave off some of those peaks, which helps everybody, not only utility.
It avoids brownouts, avoids a lot of stress within that overall market.
And so we're pretty excited about the work we're doing with a lot of our partners to bring
that to light.
What about renewable energy?
Is that an option at scale?
Absolutely.
So we operate a gig and a half of solar and wind today.
I don't think people realize that.
um The quantums of power required, I won't whistle past the core of I think your question.
You can't run, you know, a gigawatt directly data center off of wind and solar, but it's a
good mix.
And we have to use all tools in our tool belt in that solar wind make a lot of sense,
geothermal, certain markets that make sense.
So we're looking at everything, but
You know, we have very discreet ESG goals that we're going to hit and those targets our
customers do as well with some of the large hyperscalers.
That's extremely important that we're going to, you know, maintain that level of goal and
be able to execute against that.
But it's dependent upon the valley we're in now with the demand increasing so quickly.
I think a lot of us are looking at all options of power generation and being able to solve
for it.
But I do still believe and see that a lot of these
traditional types of solar and wind and other environments.
And I mean, I don't know, as a technologist, I would be remiss not to talk about
batteries, right?
The advent of batteries and the ability for us to produce them exponentially cheaper and
tucking them in the grid is very exciting to me where the battery element of that will
also start to help us, you know, be able to utilize a lot of this green energy, if you
will, throughout our portfolio.
So I'm really interested to hear your thoughts about the future, even more thoughts about
the future.
I know you spoke just a few minutes ago about what I would say maybe is a prophetic
prediction that digital reality would be the home of AI.
And so I'd like to talk about just your thoughts on what's ahead of us.
Obviously, there's just a lot.
that's disrupting, right?
Not just in this industry, but the entire world.
so what are your thoughts?
I mean, what's up ahead?
Where are we going?
What are new technologies?
Let's get into the future a bit.
No, I love it.
I love it.
um there's a lot to unpack there, right?
You can go in many directions, but I'll stay at the highest level.
with, with AI, you can view it as a disruptor and it is, it is going to be disruptive
across every vertical, right?
Like, like it depends upon how exposed, um, our audiences to AI, they're, it's a bubble.
It's kind of noise.
It's not really like it is pervasive.
It is disruptive.
And so I always try to like impress upon everybody.
Get on the right side of disruption.
Lean in, not only use the tools, because we're still very, very, very early innings.
Very early, right?
Where it's augmenting our emails, it's kind of doing some adjacent tasks.
Very excited about agentic AI, where you can have agents act on your behalf.
Also very excited about how reasoning is working, where you can have multiple models look
back at an individual prompt or question, just to ensure that it's not hallucinating.
It's those types of things that I'm super excited about which will require an exponential
amount of more of infrastructure.
And so very excited about the positive sides of what we see artificial intelligence doing
for our lives where we have a project that we were able to work on and very excited about
called Gephian.
It's in the Nordics.
It's with the Novo Nordisk Foundation.
It's the largest DGX pod in all of Europe.
monumental amount of pharmaceutical research being done there.
And so that'll directly impact our lives, being able to bring, you know, different types
of medicines and different types of capabilities in the pharmaceutical arena, but just how
AI is going to benefit, you know, society.
I would invite everybody to dig in.
So it's going to be transformative.
And I would say it's going to change not only the shape and tenor of our daily lives from
a professional perspective, but our personal lives.
And so
How I see that requiring infrastructure, digital infrastructure, more networking.
One of the things I'll explain to you is seven months ago, I wrote a blog around private
AI exchange.
I invite everybody to go read about it.
Would love welcome feedback.
But what we really representing at that time is don't overthink it.
Just like the way that cloud evolved and all the intricacies of cloud and requiring an
exchange and doing it privately and secure.
Same thing is going to have to happen with AI.
And so you're going to see more monumental amount of bandwidth required, where it's going
to be an optical interconnect.
It's all going to be about throughput, because the amount of data required for a Gentic
and some of these other AI capabilities that we're looking at two, three years out, it's
going to require a monumental amount of connectivity.
It's going to require a lot of power.
You want to do that power efficiently.
I'm very excited about how we use AI in our own infrastructure to operate more
efficiently.
watching, you know, I don't know, I always say like, don't get too inundated by it.
It's still a lot of basics where we use AI to watch our pumps, to watch our filters, just
to monitor the effectiveness of our infrastructure.
And so we've been able to save a tremendous amount of water, a tremendous amount of power.
How we see this world evolving to support AI is pretty exciting to me.
And just the way that, you know, it's going to impact our personal lives as well where
I use a lot of AI just to help me kind of track my day, track our family and like what
we're going to be planning.
I mean, I'm sure a lot of our audience today is utilizing it where, hey, I'm thinking
about traveling to this market.
The ability to tailor make an outcome for the broader, you know, professional personal
lives with AI, it's monumental, right?
Like some of the, some of the models that I've been exposed to because of our customer
base and some of the hardware that I've been exposed to.
It's transformative, right?
And so again, it can be used for good and bad.
I spent a lot of time trying to be an optimist and get on the right side of it and making
sure that it's utilized for good, but you can't hide from it.
So I tell everybody to lean in and be sure that you understand what you're buying into and
just bringing it back to co-location, the core of a lot of our conversation today.
Buyer beware and uh start to engage early.
It takes a very long time, right?
Like so...
The supply demand imbalance is monumental.
I don't see that changing anytime soon.
So if you're looking to bring something to market, you want to start to plan early, not
only to get allocation of power, but then some of the technologies behind it.
So the chips, the cooling technology is very excited about that.
We've pre procured.
We have a very large vendor manage inventory so we can help a lot of our customers
overcome some of those challenges like our partners with WWT and other suppliers where we
can help you.
not by a science project, but by an outcome.
And that's what most customers are looking for today.
So, pretty excited about how a lot of this is gonna evolve over time and quite frankly,
what it's gonna be able to produce for our lives and the longevity of it, because it's not
gonna go away.
Boy, I love your energy, I love your positivity.
is ah great.
One last question.
this is, I guess my question is, looking at the young technologists like you, like you
were, and all the young people today that are looking forward to all of this, and they're
graduating college, maybe they're working in the industry right now.
and they're thinking, where do I need to be to really uh take advantage of this shift and
not only that, make sure that my job is not disrupted and that I can really have a place
in this future, what advice would you give to them?
You know, knowledge is becoming free.
I always start there.
so, like having knowledge isn't enough anymore.
And so, you know, one of the weirdest things that I'll try to impress upon everybody is
having interpersonal skills and being creative.
Like creativity is such uh a gift and it's a hone talent on how can you work with
individuals and help them be better and help them be successful.
I think that's where I would spend a lot of my time is on the interpersonal skills.
because a lot of the tools are gonna afford you the knowledge and a lot of people will
have that knowledge, but then how it's applied and when it's applied and the creativity of
that application, that's where I would impress a lot of people to go slow.
It can be daunting.
Like I run into more senior level executives at Fortune 500 companies where like, I'm
done.
Like I don't know where this is going.
Like, oh my gosh, I got this tool, I got that tool, I hear this and I utilize this thing
and.
Like it didn't really make sense and it was like, whatever, my email was rewritten.
So what?
And I'm like, okay, you're thinking about it in the wrong frame.
So really thinking through like stair stepping through how it's going to be impactful, how
you can position yourself.
You know, we joked about it where, you know, back in 2017, when I was saying we predicted
that we were going to be the home of AI, most people didn't want to work at data centers.
The reason why I was doing a lot of that is trying to stimulate an outside intellectual
capital of humans.
to come into the industry, right?
And so how that's gonna evolve, and now it's kinda cool to be inside of data centers, data
centers are not going away.
It's a critical component of our professional and our personal lives that will continue to
be so.
So just being knowledgeable about the digital infrastructure, how that's gonna evolve over
time, and your interpersonal skills.
We all have to work with humans, good and bad, and so leveraging that tool set and honing
your skills to be able to be creative with your knowledge.
That's what it's all about and that's what it will take to be successful going forward.
So, so insightful.
Chris, how can we learn more about digital realty and possibly the open jobs available?
Yeah, no, absolutely.
So our website is always good.
um We post a tremendous amount of jobs on LinkedIn.
um Cindy Feenemann, our head of HR, um should be frustrated if I didn't reference that as
well.
But I run an ongoing blog post.
Read some of the older ones.
Like there's a full series of, know, is your data center going to be obsolete?
I go into more details around electric distribution, cooling, how you can retrofit things,
because it's not only about the three gigawatts we have, it's the
you know roughly 4 gigawatts we have coming online in the future so there's a lot of
dynamics there so go read those blog posts and be careful there's a lot of noise out there
with the braggawatt so you wanna dig into the details so that you can understand what's
real
I encourage everyone to follow Chris Sharp on LinkedIn.
I think I started following you a few months ago and I've become smarter and I agree.
Love your blog posts, love your posts.
Chris, thank you so much for coming on the show.
I've had a great time.
I appreciate it.
Thank you for the kind words.
Thank you.
Take care.
Okay, we're gonna stop.
