The UPSTACK Podcast

Welcome to the Upstack podcast, an ever evolving conversation on all things digital infrastructure, giving tech leaders food for thought as they push to stay ahead of the technology curve. I'm Alex Cole, and with my cohost and colleague, Greg Moss, we invite you to join us as we talk candidly about the latest technology infrastructure topics. Stay with topics. And and to the listener or hopefully listeners out there, we are honored to welcome Kevin Sanders as our very first guest on the Upstack podcast.

Thank you.

Hopefully, this won't

be the last podcast.

I was gonna say, hopefully, you're willing to come back. You're always welcome. As a guest in the Upstack podcast, you're always welcome to come back whenever you see fit. But, Kevin, allow me to to to share your background and some of your bona fides. So you're a principal consultant, data center strategy at EYP Mission Critical Facilities.

You've got over thirty five years of experience not only in planning and design, but also implementation and management of data centers, infosystems, network communication systems, and many other things as well. You're a

proud

grandfather as we established in our in our pre call and, frankly, an all around fascinating person. Today's episode is all about power, But maybe that's the wrong context. Energy might be more accurate.

It's energy in the context of data centers.

It's really about where is power today. Mean, the power market just seems to be everywhere. And I think a lot of people are having trouble looking at its future and what does it mean to them as businesses. So one of the first topics I'd like to kind of touch on Kevin and really hear more about is these the notion of moratoriums and they're becoming more and more commonplace in the market. Mean first Singapore, now Dublin, Ashburn seems to be on the list of an up and coming moratorium.

And I think this is posing a lot of issues for customers that need to reach these markets and will fastly need alternatives. What are their options? Is this the constant?

Yes. I mean, if you look at the context of power discussions in the sort of ten years, we never thought we would have to compete for the availability of power. In The U. S, in particular, to a slightly lesser extent in Europe, we always felt that if the power requirement was there, it was simply a matter of time for the infrastructure to be brought in. In the last couple of years, the whole issue around resiliency and the sort of green energy campaigns have put small amounts of pressure on the market.

But the growth in the use of power for data centers owned by corporations, colocation facilities, Bitcoin miners and the increase in use of electricity in other markets that we hadn't seen has put a lot of pressure on the infrastructure. And so if you look at a place like Ashburn, that went from all the power Dominion could possibly provide for a number of years to a situation where they're they're in trouble now in terms of having to supply it within the confines of what was the Ashburn market. And now moving out, outside of that market, you have a couple of, you have a couple of things you're competing with. It's not just the actual power itself or how I'm going to produce it, although that is an issue. It's the infrastructure to deliver it to these locations takes many years and chews up a lot of capital.

So even if you want to move out of a given area and develop another location, that infrastructure has to be brought in. That's a large amount of capital investment. The other thing is the market started to pivot in terms of power, particularly again to data centers. A few years ago, there was this push to let's try and find additional alternative energy supplies. So there was a move three, four years ago to gas fired turbines that would provide significant amount of power to data center campuses.

The problem with that, you know, I call those sort of, you know, political moratoriums where the use of natural gas is now becoming restricted in several locations in The US and potentially in Europe. And so that's slowing down that as a as an off ramp to providing power localized. We've also always talked about the use of wind and solar. And while those are particularly good, they do come with downsides. The wind isn't always there and the sun isn't always shining.

And if you're building a data center in particular and you want to have the kind of resiliency at a Tier three level, you have to you still have to have regular power backup capable of carrying the entire data center, which means at some point, you can't completely replace the available utility level power. And so there has been this development and discussion around the use of microgrids, which is let's use wind and or solar or other alternatives. Let's use battery storage locally. We'll use battery storage as the buffer for wind and solar over a short period of time. We'll still have diesel backup and utility as the third level.

And we'll try and achieve essentially what a Tier three standard would look like. And that development is just becoming apparent. The economics seem to be there. There are a few clients who we're working with right now who are developing pretty robust microgrid capabilities throughout The U. S.

There are some going on in Europe. So Europe, another issue, the political availability of gas and the impact that that's going to have on that in their power market is one thing. And that transcends the existing situation with Ukraine and Russia in that I think the Europeans are going to try and permanently wean themselves off of large scale single source, which means they are gonna start doing multiple development in other areas. Really interesting, a couple of weeks ago, the EU passed an advisory to the particularly in the Nordics, those locations that use district heating, if you're familiar with that. In many countries, heating isn't done within the confines of a house.

It's there's a district heating plant that either the government or private entity runs. They're saying that, data centers need to share their heat output using heat pumps back and forth between, the district heating locations as a way of beginning to come up with alternate methods of supporting large scale data center, power consumption. And so there are going to be layers added to providing power both in The U. S. And Europe.

As is always, there isn't going to be a single source solution. It's going to be layer upon layer upon layer of good engineering to allow multiple power capabilities to begin to reduce the need, and that begins to ease the requirement for moratoriums, you know, in given locations.

Kevin, we talk we talk about moratoriums, and you mentioned easing. So you do feel like this isn't for the long term. Like, moratoriums may be changed or lifted in the I think they're symptoms. Near future. Okay.

Yeah. And I think moratoriums are one of the symptoms. You'll see more. So moratoriums really follow the class of, you know, sort of putting a controls over the amount of power, traditional utility power in a given area. But there are others other than just a moratorium in that area.

The availability becomes one. The engineering requirements become another. Safety requirements become another. So moratoriums may get eased, but other requirements such as legislation requiring you to use alternate energy sources. And they can be as equally disruptive in a market because they can have an immediate impact on the cost to deliver services to the end user.

So Singapore was supposed to be a two year moratorium and now it just seems like it's an indefinite moratorium. Is there an end in sight? I mean, you talk about all these great and creative ways to kind of offset and come out of a moratorium. I mean, is this a year in the making? Is it

No, two there's really no good prediction per se. Part of it is an engineering crisis of being able again to have the transmission and the feeds into the traditional sectors in Singapore. If you understand how Singapore is laid out, there are huge concentration of data centers being built in the downtown area. They begin to they began to back off and go to the to the North and the West slightly, developing external, resource capabilities, huge amount of data center developers went into there, and then there was a return back to the downtown locations. And so there needs to be an infrastructure investment in the transmission capabilities.

But just in terms of sheer power and the energy requirements, it's becoming an issue. There are some subtle issues also about data centers being within large scale business, facilities, large, office towers and the impact that COVID had on people coming in and wanting to be in the city. So you have this back and forth geographic requirement. Do we stay out in the fringes just outside the major downtown area? Do we stay within the existing infrastructure?

That's one pressure. I just think it's the sheer engineering, and the availability of the power grid that has to catch up large scale capital investment, many years to get that done. And I think that there anyone who makes a prediction that it's going to be solved in a year or two, I think, is overly optimistic. I think it will be solved. Singaporean government is always very clever in how they handle these kinds of issues.

But I think you're conceivably looking at a what I call a three-three-one, three years to analyze and fix, three years to start building and get and then three years to roll out. So you're looking at a six to nine year period and one year of use going, it works. Now what's the next step?

So patience is what we're recommending for all those involved.

And that's going have impact on pricing because what's available is going to get locked up.

Right. It does beg the question, customers, how do companies operate in this type of environment with moratoriums in general and the time it will take to move beyond them. Greg, you touched on Singapore and how it seems like an indefinite moratorium. Kevin, you also touched on using Ashford as an example and the issue around transmission. To move to another location in the general vicinity doesn't necessarily fix the entire problem.

So do we throw our hands up in the air or how do we navigate No. This

I think curiously enough, there has never been master planning for this market before because this market hasn't put this kind of pressure on it. I think what you're gonna begin to see is public utilities commissioners in major states turning to the power providers and the data center providers and major corporations and say, we have to do some planning here. And we have to talk about how we're going to do the transmission allocation, where the investments would be. I think you're going to begin to see some tax relief coming from various states into particular sectors. If you look at if you go out from Ashburn, you begin to get into the locations that were old scale manufacturing facilities that were have been abandoned and they have significant, not huge, but significant amounts of power, which were brought to those facilities and it's great to get those reuse, but it's not enough and it's not clustered correctly.

And so there needs to be some the old phrase urban planning done on how we're going to reallocate that and the impact it's going to have to rate payers. And I think you're also going to see some changes in the way rates are done at the state level, allowing those tax advantages not to just impact all rate payers, meaning consumers, but begin to impact the commercial sector in a greater fashion. If you're a corporation or a user, interestingly enough, the place to start is not at the data center. The place to start is your application portfolio and what you're doing. I always tell our clients the best engineering is to make sure you're not doing what you did for years, which is, oh, jeez.

Let's just buy some more servers, slam them in. Who cares about doing application control? And I think the whole cloud, universe helped them because cloud, quite frankly, was expensive if you used it incorrectly. And I think a lot of corporations and a lot of private individuals realized that they had to be pretty judicious in their use of applications and communications back and forth to the cloud and the use of data. And I think that taught institutions to be clever with their app portfolios.

The first thing is consolidate, and particularly for large corporations who I don't will still build data centers, but not of the scale and those sort of wanting capacities that they did over many years. They use a phrase that I call data center sourcing, meaning how much goes to the own data center, how much goes to what I call an intermediate data center, meaning it's a colocation facility that has a cloud back end, either public or private, and then how much goes to cloud cloud. And I need to move my portfolios and understand that growth rate in that period of time. And it is amazing how much you can reduce the need for your data center load if you do that. And it's akin if gasoline gets expensive, plan your trips.

And that's what they need do with the app portfolios. The second issue is make a decision on the level of resiliency needed. The U. S. Sort of grew up and Europe followed with the Uptime Institute Tier three, N plus one is the standard for resiliency.

And it's a great standard. I'm not knocking it. It's there. But for some users, it's not required. And that chews up a lot of power when you do that.

I still believe that there's a growing market, particularly for the small users, 50 racks, 100 rack requirements going into colo to use less than that standard, potentially repurposing older data centers right now without having to make that hard scale capital input. And coming up with the data center with a lower resiliency and using good standard disaster recovery, Doctor capabilities by making sure your data is replicated someplace else. We do this for clients now. When we're helping them with data center sourcing, one of the first things we do is say, let's take off the table anything which does not have to be in a very high end data center environment and let's move it to other rational acceptable environments. And it's not so much that it lowers power, I don't need the duplication of power, the duplication of the physical capital assets at both locations and the numbers of disk spinning and servers running.

So what I'm hearing Kevin is by reducing the levels of redundancy needed in particular use cases and then kind of refining your own internal applications to use less power, be more efficient is really an answer to the rising power It's costs

one. Again, the solutions are always multiple. I make it akin to the 65 year old guy who goes to the doctor for the first time in ten years. It's never just, I have an elbow problem. You cure yourself through multiple layers of change and that's one of the layers of change that we just talked about.

Buying smart in the market, moving to locations that are potentially less convenient. So there are data centers going into places with cheaper power. States we would not normally think of, North and South Dakota, Wyoming, Montana. We all know about Pac Northwest. So you can go seek cheaper power, sometimes cleaner power, rather than what everyone put your data centers in this in what I call the NFL cities.

The major locations I can fly into and fly out of in a convenient space. And that distributes the power requirements incidentally. We have many clients right now who are trying to stand up data centers in these more remote secondary locations that have some power availability, slightly less convenient to get into, power can be cheaper. And the challenge there is just make sure there's robust enough network to be able to move the data back and forth.

So that leads me to the next point, network. Right? As you get into these more remote locations, the challenge becomes do I have access? And access to 99% of the companies today is important. May not be latency sensitive, but it's still an important factor.

Do you see the network infrastructure within The United States and other big major markets around the world moving along with this plan to help offset the need for such expensive power?

I don't know if there's coordination between those two efforts. What I do know is that mile for mile, it's cheaper to bring in dark fiber to a location theoretically that to bring in hundreds of megawatts worth of power. And so I have clients right now in some of those states that I just mentioned, who are in secondary cities, who have very, very good capabilities network within the confines of that secondary city and are willing to set things up several miles outside of that city using cheaper power or building cogeneration capabilities and spend the money to bring the network speeds in. I think that network companies are more willing to do that kind of investment if they know they have an anchor. So if you're put a consortium together and you're going to be putting in 100 megawatts worth of campus computing, 200 megawatts over the course of multiple years and they know the market is there, you need to do early on is bring in those network providers and get them to come along for the ride and provide it.

But yes, I tend to find that it's easier to solve that networking issue. But again, we get back to the app portfolio. Don't put your applications that require molecular modeling, rotating proteins, if you're a pharma in real time. If you're doing transaction processing and you're trying to chase an investment market, it's probably not the place to go. But if you're an average user storing data and you have low use so you can sustain the kind of millisecond delay, it's a perfect market for you.

So what I'm hearing, gentlemen, is there there are challenges mounting, but there still is a path forward. And maybe it requires a bit of flexibility and nimbleness, but, you know, businesses can still be created, work can still be done.

Yep.

It's gonna be okay. It might require working with the right partner, expanding this the the list of suppliers you might be willing to consider as well as geography always playing a factor

as well.

I've noticed, you know, and it seems like what Kevin's saying is this can become a competitive advantage for some companies, right? To be able to reduce your costs while your competitors are the competitors have rising costs. I've noticed, you know, most companies have a focus, right? And their focus is their business and their core competencies. It's very difficult to look outside and become an expert at everything.

So, you know, I definitely feel that third parties would be extremely resourceful when trying to analyze the future for businesses.

I mean, one of the things that, again, one of the major things that we do is provide long term strategies like this for our clients. And it always amazes me that more companies don't recognize that a data center strategy is a pretty integrated thing into a company, has a huge impact to OpEx and CapEx long term, if not done correctly, and to competitiveness. And they don't plan the data center side of their business strategies with the same kind of skill to match their business strategy. I was very concerned years ago when everyone did this great push to the edge. Well, from a data center strategy standpoint, that presents some challenge and some network issues.

And there wasn't a lot of follow through in terms of what else do I need to drag out to the edge, okay, to be able to match that that business philosophy. For large multinational corporations, it's extremely important right now. You know, a lot of companies grew up having multiple data centers on multiple continents and are now looking to consolidate on multiple continents in doing the three legged stool. You know, major location in Europe, major US, major Asia Pac, list link them all together, a couple of Doctor locations at each of those on the continent, share critical data back and forth when legal and when necessary, and get out of having twenty, thirty locations in Europe doing small scale data work, same thing in The U. S, same thing in Asia Pac and link them together.

That sounds like an easy thing to do, but it's not. That's a multiyear project, takes a significant amount of planning to do. And so again, your ability to take advantage of power cost, power availability is really based on an entire project, which needs to be thought out so that you can follow that. So it's one thing to talk about that strategy, another thing to drag the rest of the company with you and in fact, make that happen.

And that's very interesting, but you're bringing up a very specific use case and a very specific type of company. These are the larger, let's call Fortune 500 companies. The last decade has been around agility and the ability for these high-tech or new innovative companies to be agile and spin stuff up on a So moment's how do you merge those two, right? How do you get the information you need as an organization to be efficient quickly?

Well, I think largely the smaller guys have taken advantage of the cloud explosion over the last couple of years because it works really well for them. The scale of everything is smaller. They can move around. They can take advantage of smaller colocation opportunities in multiple locations. They've resisted building their own data centers.

There was a period of time when we saw small companies that were 50 to 100 rack requirements essentially wanting to build out their infrastructure in a highly planned environment in a colocation facility. They've mostly given up on that and they're adopting, the backdoor, which is put something in a colo, have a backdoor out to a cloud and then manage that valve between what I have in a public private cloud and then what I need to retain within a colo is my own private, you know, work. And so it it can be easier for them. They, however, being faced with longer term contract requirements. It used to be that a lot of colos were willing to sign contracts for small users for twelve months or less.

That's becoming more difficult as the amount of colocation facility available space begins to dry up in particular markets. They are looking for longer term contracts. And so the ability to jump, gets lessened. And that's where I said, I think this, sort of barbell market needs to flourish, where you have the on one end, the traditional colocation capability with backdoor up to cloud. On the other end, you need to have good, stable providers of smaller, less expensive, slightly less resilient facilities at a lower cost, maybe in secondary locations to try and diversify that market.

So they have a place in fact to turn to.

Interesting. I'm seeing the panic set in. I'm seeing companies get concerned, right? I mean, where's my infrastructure going to sit if there's no power left in a particular facility and they're pre buying power. So they're buying power that they're not going to use today.

They're paying for it. Do you think this is a smart move or do you think it's, something they should sit on?

That's a that's well, if you're in a market that's prone to a moratorium, you have a limitation on pre buying power, you know, putting money down. At the end of the day, there's lots of people who need to make money off of it off of that power. Putting a a hold or, you know, right of first refusal down is a is useful, but long term, that's probably not, a given solution. I have clients who are, in fact, buying lots of colocation facility, and reserving a lot of power with it, making a significant capital investment just to have that availability over three to five years. And they're willing to spend it because it's still, if you do the economics, it's cheaper than building something or rebuilding one of their existing data centers.

They can do it smaller folks a lot less. You know, I think that the for them, you know, the hyperscale build market is one of the things which has spiked this issue. It was an explosion in the last five years. It chewed up an enormous amount of power. Most of the colo providers, both to the existing market and the traditional investment colo market turned to hyperscale as a way to, relatively fast return, albeit thin margins, over a long period of time, chewed up a lot of power.

And unfortunately, a lot of those designs were good, not great, mid level PUEs. And there's an entire reserve of stuff that's that is built and is being built, which serves that market, which controls a lot of power. And it's gonna be interesting to see what happens with that within a five year period, whether or not there's a reengineering of those facilities to bring in alternate power as a way of bypassing the moratorium. And that's that sort of microgrid that I was talking about, assuming they have the space and they can get the the permitting for that. Or doing major reengineering of some of the sites that are already five years old, that in five years will be ten years old and could be made significantly more efficient, more effective in the use of the existing power.

And I think that engineering alternative is something that's gonna become a necessity in the in the in the coming years. We simply can't build a lot of good facilities. We have to build better, more highly engineered facilities, which make significantly better use of available power.

When we look into the future, I mean, as we've mentioned, challenges rising, but we know there's a path forward. Looking at the future, panic is on the rise as well, Greg. So you've got clients coming to you saying, what am I doing here? What is the first step you recommend a client take given the current environment and what the future looks like?

I mean, my opinion is pick up, you know, so it's back to the understanding that you're good at what you're good at. I think you got to look outside of your organization to really understand not just the global landscape, but to understand what this means to your organization, how do I become more efficient, how do I position myself for scale, right, not just today and not just two years from now, but ten years from now, right, predictability. And there's people out there, I mean, not to plug the company, but like Upstack and EYP that could help you better understand what that means to your business.

Definitely plug the company, Greg. Good job.

Upstack, call us.

Kevin, it's our podcast. We can do what we want. Okay? It's our podcast.

This is what I call turn now we're turning the AC up here, Kevin.

Turn the AC up. You you make sure I mean, where to begin? I mean, the

landscape It is a good point.

With pitfalls. Yeah.

Most people have a knee jerk reaction. And usually, that's a that's a bad outcome in a in a short period of time. It's looking at your requirements when you know, panic can be a good thing because it forces people to have a higher degree of willingness to accept change and do, more analysis than they normally would. When power and colo when colocation was available and relatively inexpensive and not terribly competitive in terms of getting space and power was known and controlled and no moratoriums, it was an easy knee jerk reaction. Let's just move to the next large market and acquire it.

As that becomes difficult, power availability, all the things we just talked about, Power costs are going to rise. Availability of colocation space in general has going to have challenges, particularly some types of the resiliency levels that people are looking at. You have to have a more planful approach to this. And I always say there are three things you need to look at in terms of your strategy. Your long term risk, what impact is this new formula going to have on your business model in terms of your profitability, the markets, how quickly you can move, availability of your OpEx and CapEx.

And that's the second issue, your financial line. You need to look at this in terms of what is the impact, the cash flows and what are the impact to your profitability over a period of time because strategies impact your bottom line. You have to be able to monetize a change in strategy and a change in your infrastructure costs to what your profitability looks like. And you might have multiple methods of solving the problem. I always say scenario analysis is a great thing.

There are generally two, three ways to solve these kinds of knee jerk problems. Look at them, analyze them within some degree of reason, understand the impact of your economics over a five potentially longer year period, choose one that makes sense and that loopback cycle, if that gets you back to having to change your business model slightly, understand where you're going to push your increase in cost to. Are you going to lengthen your recovery timeframe for your capital investment? Are you going to have to pass some of that on to your clients? It's there.

It's mathematical equation, a little bit of science involved here, but it's homework that your teacher used to give you that we all never wanted to do. It's homework time again. You got to do it. Now

I'm starting to sweat with the mention

of homework.

With challenges rising as we know, often there's opportunities to be had. So if you know where to look and maybe you know who to work with, those opportunities can be taken advantage of for the long term benefit of your business. So well said, Kevin. What what is the first step, you know, a client and a customer should take? I mean, given all that's going on in the market and the changing landscape, a knee jerk reaction is possible.

Where do they begin? What's the right first move?

Well, first move is not to let it get the knee jerk, right? And I think that this is a real issue unless you're on the inside, you're not seeing it unfold and it's something that everyone needs to pay attention to. Rising power costs, availability of power is as real as it's ever going to be. Start to look inside your application, start to look inside your application stacks, data center environments and figure out how you can be more efficient, how you can build for scale and we're here for you.

Kevin, we we could go on clearly, but we also recognize you have a day job. So we won't give away all the secrets on today's episode, but we do appreciate you being with us and truly our very first guest our very first guest on the Upstack podcast. So we very much look forward to welcoming you back Yeah. In the coming months. Hopefully, we're still around, Greg.

We'll see.

You're appreciated, Kevin. You're appreciated.

Thank you. I had a lot of fun, and remember, we're all in this together.

We are we indeed are all in this together, and it truly is amazing to share knowledge and to speak to the power of the the industry and, we think, the services we provide. So until next time, everybody. This is the Upstack podcast. Thank you for listening to the Upstack podcast. Don't forget to like or subscribe to the show wherever you get your podcasts.

We'll see you next time.