Ep. 189 Ken Sullivan, Chief Executive Officer at Bay Compute | Data Center Go-to-Market Podcast

Speaker A: I think the idea that we're over building is going to be so wrong. Um, and there are plenty of people who don't think that we're overbuilding and think there's a lot of narratives out there that we are, um, and we need to have data center moratoriums and all of this. I just think that the uh, proliferation of not just large language models and small language models, but robotics and the once, once we have this proliferation across not just your early adopters of tech, but your late adopters and the next generation growing up, being AI native, as, uh, they continue to build companies and build products, I think that there's no shortage of demand that is going to exist for the next decade. Um, so that's one thing that I think is often talked about, as you know, AI being hype, AI being a bubble. I think we are completely underselling. You look at China, they're building 100 gigawatts of nuclear. I think we could be doing the same, but because of regulation we aren't. So I'd say that is one thing that I think is completely underestimated and missing the mark is how much demand actually exists out there. Hi, this is Ken Sullivan from Bay Compute and you're watching the Data Center Go to Market podcast.

Speaker B: Hi, it's Joshua Feinberg, host of the Data Center Go to Market podcast. Today I'm welcoming Ken Sullivan, who is the CEO of Bay Compute. Ken is based in the greater New York City area, and Bay Compute's tackling a really difficult challenge. Convincing data center operators to buy software to optimize and manage capability, what they call data center orchestration. In an industry that traditionally solved capacity problems with spreadsheets, dsm, um, tools, data center infrastructure management, sometimes over provisioning, sometimes infrastructure upgrades, or simply buying more capacity. Their positioning. Bay Compute's positioning appears to center around synchronizing data center assets and power systems, cooling systems and Compute resources to unlock stranded capacity and improve utilization. Ken, welcome to the podcast.

Speaker A: Thanks so much, Joshua. Great to be here.

Speaker B: It's uh, great to have you here. And before we get into Bay Compute, I'm really curious about your personal journey as a founder. How did you arrive in your current role and what problems along the way earlier on in your career convinced you that Bay Compute needs to exist?

Speaker A: Yeah, so I spent, uh, time building data centers in the US Marine Corps as an intelligence officer. And when we got the first AI tool in the entire US Military, I was dealing with power bottlenecks from the four deployed locations where all of our different Centers were existing and really taking manual efforts, shutting down H vac units, using daisy chain generators to keep uptime on the systems during critical operations. And the mitigation efforts that I was using very rudimentary software. I ended up meeting a brilliant co founder out of MIT five years later and he ken, I wrote software that automates a lot of those, uh, manual methods that you did. So what we did was we then took a look at the data center industry, really following the electron from the grid all the way to the chip, looking at where is there an area of the market that isn't being serviced currently? Because there are bottlenecks all along the power chain and inefficiencies all along the power chain. And really what it came down to was where is there an area that isn't being serviced? And it's really at the data center itself. Uh, particularly in colocation environments where they don't have the same tools that a hyperscaler builds in house. And they really have never been incentivized to make these tools themselves because, uh, COLO has always been someone that is structured like a real estate company. It's typically a REIT in the management level and they act like a real estate company when they have filled up their, when they've leased all their contracts, just like filling up a building in real estate, they go and get a new building. Well, we saw that coming to a head two years ago when we started exploring this environment. And it was going to become very difficult for data centers to continue to build new buildings because of the backlog with power. We're now seeing anywhere from five to 10 years of grid interconnection queues. And then you're getting, if you're going off grid, you're going behind the meter, you're running into supply chain issues. So we really started looking at how do we deliver this software so that they can optimize the power envelope that they already have. There's a lot of power that gets delivered that isn't used most of the month. Especially when we look at AI, uh, inference workloads, they spike much higher than the average power utilization. And what we can do is then help them to mitigate those spikes during these peak power events and thereby free up more sellable capacity.

Speaker B: Yeah, it's really interesting how larger environments, hyperscale, wholesale data center providers and for sure mission critical and enterprises and military had to figure this out early on. But colocation a lot of times is optimizing for. You're right, they're optimizing for A lot of real estate outcomes. I often joke with that. If you ever get a friend or family member that wants to know what colocation is, tell them that it's like co working. It's like we work for servers. They can buy a membership, they can bring all their servers, they can bring all the racks and they live there. And the, the co location providers that end up having much richer margins usually end up moving upstream and trying to position themselves more like IT services companies. But yeah, it's a, it's a mature model facing some really big challenges now around power utilization.

Speaker A: Yeah, that's exactly right. And you know, I think the WeWork example is spot on. And when you look at what the COLO offers, it's pretty basic bare bones services. They basically have an agreement with their customer. Hey, we're going to offer you space, we're going to offer you cooling in between these temperature ranges and we're not going to know what you're doing. Uh, WeWork doesn't know what the its uh, entrepreneurs are doing in its environment or its companies are doing in its environment. In the same way colos don't know what's happening with the tenant workloads that are running in their environments. That's, I think a big differentiation between what we do and uh, some of the others in the market are targeting is we don't require that workload visibility. So we can work with colos, we can work with any data center whether or not they have visibility into the workloads. We can do the compute orchestration that my co founder was writing about five years ago at mit. And that is something that we can do, but it's not a requirement. We can still look at the IT power draw in order to inform the other aspects of the data center that we can then use to create that capacity.

Speaker B: Yeah, it's fascinating and it's really cool to see you looking for creative solutions to help them make the most of the power they actually do have. I'm sure you get to talk with a lot of infrastructure executives on a really regular basis. So I'm really curious to get your perspective on what you see as the biggest go to market challenges that infrastructure companies are facing right now.

Speaker A: Yeah, so the biggest challenge we see is kind of the getting them to frame the problem correctly because a lot of times it is framed as an energy problem. Right. I can't get enough power when you really need to frame it as a revenue problem. So when you look at the ability to unlock capacity from what you're already paying for that is a revenue increase rather than just a I'm out of power and I can't do anything with it. Um, so what we can provide is really that ability to increase that revenue for the customer rather than just looking at, okay, I can't get enough power. Is there an additional power source? There's more you can do with the power you have rather than trying to go behind the meter with solar plus batteries, with natural gas, with all these other, uh, or maybe getting additional um, power From PJM or etc. Wherever your data center is located. So lots of creative ways to increase their ability to increase revenue.

Speaker B: Yeah, that's really an enlightened perspective, especially in the US in many ways. A lot of sustainability experts were talking about this in Europe for 3, 4, 5 years because of directives over there. But it's become a lot more real in the last two or three years because of the high density AI workloads. If Jensen Wong and uh, Sam Altman had not become celebrities three years ago and we hadn't seen this massive adoption curve on generative AI, it's possible that this wouldn't have happened so quickly in the US but here we are. From your perspective, how have you seen AI change your strategy in the last year?

Speaker A: Yeah, so the, the biggest thing is the, the density I would say. Uh, so what we're seeing is the construction really shifting from a mix of air cooled rear door heat exchanger, liquid cooled to now everything has to be liquid cool because you just can't have, you just can't move it enough heat. The thermodynamics break down um, at the densities that the new racks are coming in at. Uh, so what that means is you have just a different makeup of all the data centers that are being built going forward. Uh, you know, whether that's direct to chip, whether that's uh, plate, um, or liquid cooled. And so what that means is you have, with these different architectures, everything is constantly changing. And by the time you sign your vendor agreements, if you're building a data center, uh, or rather from the time you sign them until the time it's actually built, there's better technology on the market. But I think one of the things that is misunderstood in the environment now is that there's still a need for that inference. Um, we see the spot price of H1 hundreds is back to where it was three years ago. And that's just because there is so much demand. So even though you have technology that is being built so fast that there's almost a Question of like, should I wait for the next generation to come out before I build mine? No, uh, you should absolutely build with the latest technology. Understand that it's going to be not obsolete, but it's not going to be the cutting edge in six months. And that's fine because there's still going to be demand for your technology, for your services.

Speaker B: Yeah, I hear that a lot across the different parts of data center tech and facilities and construction real estate. Among the operators is figuring out these timing differences and get every, getting everything to match up correctly because there's just such a massive amount of capital required for these projects. There's so many supply chain constraints not just around utility power, but around the raw materials. There's supply chain constraints around a lot of the hardware that goes into the data centers. You're in a really interesting category though, orchestration and helping all of this happen, helping them get better utilization out of their power. But when clients don't even know that they need your category, a lot of people always talk about trying to be more client centric, meeting their clients where they actually are. How do you educate the marketplace before a sales conversation takes place if they don't yet know to come look for you?

Speaker A: Yeah, and that's a major challenge, right, uh, when you're defining a new category. Uh, what we have learned is that, you know, buyer education in this industry happens through peer credibility, not content marketing. So it is going to the people you know and trust. We've been in this industry a number of years and using that word of mouth and talking to the folks that are super important and super bought in on what you're doing. Because at the end of the day, operators, the ones we're talking to, don't read white papers. They talk to each other. That's where they're getting their information. These folks all have swapped around from data center company to data center company throughout their career. And so they're talking to old friends who they cut their teeth with early on saying, hey, we're looking at this orchestration software. I think it would really help, uh, the footprint that you have and that seems to be a good driving force.

Speaker B: So you're finding that events, associations, peer communications is doing much better for you than thought leadership and technical content analyst relationships. Customer stories.

Speaker A: Yeah, that's exactly right. And you need the proof points with the customer stories and the case studies, but you really need to expose your network to what you're doing and then build from there. And the case studies and customer stories are just a supplement to what the actual conversation is which is your network talking to each other.

Speaker B: So as you're building Big Compute, has there been anything along the way that you originally outsourced, you originally delegated to a third party company that didn't especially work well, that you wouldn't do that way again?

Speaker A: Yeah, there was some initial push towards, you know, bringing on, uh, additional advisors and outsourcing some of the initial sales. Um, I think nothing, uh, nothing works better than founder led sales, especially when you are defining a new category because people don't understand it. So you really just need to get to those events, you need to talk to those people, um, rather than having a proxy do it for you because you need to be in the loop with what the customers are saying and then what the product is going to look like because we invented the technology a long time ago. And then it's just a matter of what does the customer want to see, how do they want to see these models delivered to them. And all of that has to be founder led because you understand the problem better than you know, someone on the sales side or the advisor going to, at least at the early stages.

Speaker B: Yeah, it sounds like the basics of getting to product market fit and of knowing exactly what the packaging, products and services look like, uh, who exactly the buyers are, the price points, the terms, the duration and yeah, it's super hard for an outsider to figure that out early on. There's so much interest in scaling. But the challenge is so many early clients, so many early adopters are buying you as the founder. Right. They're buying into your vision, they're buying into the, you bring into the table before you build the playbook on trying to train and clone yourself.

Speaker A: That's exactly right. And that's just something that you can't press the hyperscale button on. Um, you can't really. It's like a critical skills operator in the US Military, special forces. Somewhere I came from is a special operations world where these people aren't mass produced. Right. And it's the same way at early stages of a company is you have to take all of those ideas that you are getting from customers and really spend time with the engineering team and with the customers in order to craft and hone that product rather than just trying to mass produce it. Because at the end of the day you're going to have things fall through the cracks if you're trying to do that.

Speaker B: And actually it was a big trend in SaaS startups, uh, about 10 years ago in the mid to late 2000s where a lot of people like the ex, the early leadership team from companies like HubSpot were trying to figure out like what the perfect metrics are that we were watching. That was a leading indicator that the adoption was really strong. And then using that as a proxy for trying to figure out when you had go to market fit before you significantly hired sales. Because a lot of times like companies, early stage startups, they go and they do a funding round and then all of a sudden the new board members go and say, hey ken, go hire 40 salespeople in the next year. And like you're trying to go from having two people or no salespeople to 40 and is it even remotely realistic? And you just burn a lot of capital before it's free. So figuring out the timing on that is huge. I don't think that much of that methodology has really made it through to digital infrastructure, but it's a lot of the same challenges.

Speaker A: Yeah, it definitely is. And there's no, you know, perfect playbook for early stage, uh, startups. And it's way more of an art than it is a science figuring out, okay, what is replicatable. And it's only replicatable once you have tried and iterated and really have something that the customer is pulling you for, right? It's that they have that hair on problem, hair on fire problem that you are solving and the motion that you used to solve it for them becomes repeatable. Then you can start scaling the team and bringing on others to replicate that motion. Once you have captured all those lessons learned and you can't exactly to your point, you can't just uh, hire 40 people at once. You have to make sure that it is replicatable and you can scale the team in a responsible way that allows them to learn at the right pace to be able to deliver that product. Because what you don't want to do is have a bunch of deliveries that fall short of what you're actually promising because you have brought on too many folks and not trained them enough.

Speaker B: I see when I talk to so many founders too as well. Both in the data center industry and this has been something in the broader tech industry for a while is those early hires. It's so different to sell in a company that's building a brand that is not well known in the industry compared to a brand that's been around for decades, where there's hundreds or thousands of salespeople in the playbook is on version 117 and like it's the same thing like decades ago where everyone Said nobody ever gets fired for buying IBM. Like there's a inertia. There's a perception that there's a safe choice in hiring people that were account executives or account managers for Digital Realty, from Equinix, from Schneider Electric Converter. There's huge companies in this category with the challenges. If they weren't there super early, they may not have ever had to contend with being one of three salespeople in a company that has a part time marketer, that has a product manager. This stretch thin where they're still trying to figure out a lot of what product market fit and what the pricing and packaging and sales process looks like.

Speaker A: Exactly. And that's where all this comes back to just being an art, not a science. Because those folks do end up having a large Rolodex of people in the industry that they know. But to your point, they haven't had the experience of selling something when they were one of three sales folks. Right. They were the first AE maybe. So it is a balance of all of those things. It's also a balance of industry experts and then young hungry people who know the tools that exist on the market today that can really leverage the knowledge of someone who's already existed in the industry for a decade or two. And all of that, when you put it all together and just have good culture within the company, that's how you lead to success.

Speaker B: So I know you get to speak with a lot of data center operators, colocation providers and people across the different sectors of data center tech and data center facilities. What do you see as a go to market strategy that is now starting to fail, that a lot of people are in denial about that are still continuing to aggressively invest in, even though you're seeing clear signs that it's seen its better days?

Speaker A: Well, I think the sustainability or energy efficiency as a value proposal has really fallen on its head. Um, so that would be, that would be one area. But specifically within energy efficiency would be your point solutions. So you know, if you want to optimize the, the H vac, right, you want uh, denser rack or better H vac or better individual point solution. What you see is that when you look across the entire facility and you stack these optimization techniques, you're not getting the same increase as if you optimized across the whole facility. And so uh, the comparison that I, that I make is to a soccer team that has two good players versus one that passes the ball really well. They're probably the team that passes and defends and scores well is probably going to end up winning the game rather than just having two all Stars. Uh, and so these individual point solutions, those are coming out of a lot of the, uh, your OEMs that make these components. But what you really need is something that looks across the entire stack. Uh, because at the end of the day, energy efficiency isn't just about cost savings, it's about revenue enablement. And so that's the approach that we take and that's why we see the success that we have.

Speaker B: Yeah, it's really interesting with the downside of the point solutions and the value of the single pane of glass and integration and seeing everything together. When you brought up the soccer analogy, I was thinking of the baseball analogies too, where you'll, when you see a post game interview, you'll never see a player that's getting congratulated on having a lot of, on having a lot of hits, cycle, high batting average, low era, whatever it is, take solace and being awesome individually when their team didn't win. It's probably like the basics of PR training that they get during spring training is always, always, always optimized for your team winning first and you winning the honors yourself second. So yeah, it's a lot of figuring out how to bust the silos that prevent these companies from getting to the, the outcome that they think they really want to get to where people are fighting for their own turf, but not don't necessarily have their company's best interest or their client's best interest at heart.

Speaker A: Exactly. And especially in the colocation market, where if you're bringing a point solution that is energy efficiency, it's likely cost savings, which is likely just a reduction of power cost, which is likely passed on, um, directly to the tenants. So the colors just do not care. It doesn't move the needle for them. Um, as opposed to bringing, hey, here's a solution that is revenue enablement. It's going to allow you to sell more capacity to your tenants. That is what they're interested in.

Speaker B: Yeah, that makes a ton of sense. Most data center GTM teams are flying completely blind. 83% of your buyer's journey is happening before they even speak with someone from your sales team. The Data Center Go to Market podcast is powered by DCSMI M. We've studied over 1900 industry leaders to build a diagnostic flow framework that identifies exactly why deals fall apart and revenue stalls. Stop guessing and start benchmarking. Subscribe now to our Weekly data center GTM briefing at. Ah, www.dcsmi m.com briefing again that's dcsmi m.com forward slash briefing. B R I E F, I N M G. So you mentioned before that what's been super helpful for buyer education has been peer validation and getting the peer conversations going. Why do you think that technical content fails to generate revenue for so many companies?

Speaker A: Yeah, the technical content is definitely helpful, but that's typically not where the buyer is. Right. And so technical content typically falls short of showing an ROI and showing a number. So when we have content that we are showing, it is, okay, here's the math, here's how it worked. But that's on pages two through six because page one has the ROI up front. Here is exactly what the economic numbers look like when you use this solution. That is what moves the needle, not necessarily the technical content. That is, uh, you know, interesting to the engineers. But they're not, at the end of the day, the ones paying for this software. It's whoever owns the go to market side. Whoever owns it's typically, you know, that cfo, coo, someone in the C suite, uh, even ahead of sales, might be below where the buying power is for this type of tool. And you need to have the ROI as part of that delivery piece.

Speaker B: It almost sounds like they're trying to do too much at once with a single campaign, with a single piece of content. Similar to someone speaking at like a general session or a general keynote at an event, versus having a much easier time optimizing for the CTO track or the engineering track or the finance track.

Speaker A: Yeah, I'd say that's pretty spot on. Uh, you want to be targeted with your information. And if you have technical content, leave it for the technical audience. Um, what we find is that it's often, uh, the C suite level that is interested in this type of solution. They are the ones who have the influence over both the sales side and the engineering side. And the engineering side is the one that you need to implement it so you can talk to the sales side. They are all about it. They are so excited. Hey, my revenue's capped right now. My, uh, my sales are capped because I have no more power to sell and you're giving me power, so I'm excited. But then the operators on the engineering side, they're the ones who have to implement it. And so it's figuring out the way to separate those narratives. It's a totally separate message to the operator because at the end of the day, it allows him to run his job a lot easier and that ability to operate within the Confines of the SLAs of the customers because we put those stops in to the software that allows him to do his job easier. But that's a much different narrative than what you're telling the sales side, which is we're creating 10 to 20% more sellable capacity for you. So the intersection of those two is up at the C suite level. That's who's buying it and that's where that message needs to go. But if you're having that, ah, technical discussion, it's often left on the technical side.

Speaker B: Really interesting too to hear you bring up a couple times the value of revenue, the value of sales leaders being an internal champion for your initiative. I don't hear that all that often among companies that are building solutions for data center facilities, for data center technologists. So really fascinating to see that being part of your overall messaging and your overall go to market approach.

Speaker A: Yeah, well, I mean it hasn't been an issue right until probably the last year or so where up until then, you know, there was always another building to be built, there was always more power to be acquired. Now they are truly feeling the squeeze where, you know, their bonus is tied to additional sales and they can't sell anymore because the company doesn't have more capacity to build. So when we introduce this solution, then they get excited. So it has been a good way in the door in some organizations, but again it's not typically the buyer. So it is just a way in the door. And that's super important is, you know, mapping out who the buyer is before you even meet the company, um, and figuring out your route to that buyer. So whether you start on the sales side or the engineering side or a warm intro to the C suite level, which is probably preferred, it's important to understand who that buyer is.

Speaker B: What do you feel sales teams are consistently misunderstanding about how modern buyers today are evaluating their different options. Where do you see sales teams struggling the most with the buyers of today as opposed to the buyers from five or 10 years ago?

Speaker A: So I think, um, at least in the market that I'm in a lot are, I think misunderstanding who the um, buyer is in terms of the technical capability and how that influences the segment of the market you're going after. So I go back to the compute orchestration and that is a very interesting way of creating more capacity because a lot of data centers are running it 30 to 50% average power utilization, but GPUs can be as low as 5 to 30% power utilization, super low utilization. And so if you can increase that, which There's a lot of technology out there that exists to do that, namely ours that uh, you know, my co founder built five years ago. But we saw that as we looked across the market, we saw that not as being the linchpin for the success of this company. Because if you require that compute orchestration, you're segmenting yourself away from colos and into an area of the market that typically already has internal tools for this, uh, that being hyperscaler, someone who owns their compute and owns that entire power flow within their building. So I'd say that that's like probably the biggest mistake um, that I've seen is going after, you know, the big hyperscalers who tend to, at least some of them tend to tend to just build tools in house rather than take on outside vendors. Um, and you know, we really kept it in a segment of the market that where we can go after them, we can do that compute orchestration, but we don't have to. And that allows us to, to service who we think is really in dire need of this type of technology.

Speaker B: So it sounds like a lot of dialing in the product market fit and really understanding that ideal client profile, getting really tight on the addressable market. Like what size, what kind of scale does a uh, colocation provider need to be operating at to be a good fit for what you do? At Bay Compute is a solo single location co location operator that's doing a couple of million a year in sales. Dozen or two dozen employees. Are they big enough or what kind of scale is typically in terms of power? Do they need to have to make it worthwhile to look at a platform like this?

Speaker A: Yeah, so the answer to that is just our, our platform is infinitely scalable, so it can be, you know, one to two megawatt facility. There's a lot that we can unpack there, uh, and unlock there because it's typically very, uh, you know, those are your edge facilities that are typically a couple of years, if not a couple of decades old and then it scales up to hundreds, uh, megawatts up to a gigawatt scale because it's just on prem software. Um, and so the ability for us to flex from a small site up to a large site is all just based on the scaling of software. And uh, what we find is the customers we're working with often have multiple sites and often have small to medium to large size colos, uh, across their entire portfolio. And so it's just a matter of starting out with one site and then expanding to the others. And we're size agnostic, it's not going to uh, you know, have any, we're not going to be precluded from running in a smaller facility, uh, or a large facility.

Speaker B: Have you found that your decision maker changes depending on the size of the account you're targeting? Is there a certain size where a CTO or COO or someone like that is empowered to truly make this happen? Or with the size of colocation operators, even with a wide range, are you still finding this is going all the way to the CEO level?

Speaker A: Yeah, there is an inflection point where it doesn't need to go to the CEO. Um, and I will say from the folks we've worked with, a lot of times the CTO in your smaller companies, it's often a couple of co founders, CEO and CTO that have a small team and then it's one phone call, they're interested, both of those guys are on the call and then we're off and running. Running. Um, in some of your larger orgs it's obviously going to move a little slower, in which case you, you're probably not going all the way up to the CEO. It's going to be someone at a GM level or at maybe a COO or CFO level that needs to bless off on this project. Um, but that inflection point yet tends to be on your larger colos that uh, where it's not necessarily going all the way up, but I end up meeting those. The C suite anyway, roughly where do

Speaker B: you see that inflection point usually happening?

Speaker A: Uh, I'd say when you're getting into multiple hundreds of megawatts, uh, that's when multiple facilities that are in the hundreds of megawatts. That's when you're looking at, you know, someone who has more priorities because they're often building the next one and acquiring and all of those things. Uh, so it is, you know, in your 10 to 20, even 40 megawatt might be a CEO CTO conversation. And then when you're above that, uh, you know, a 20 plus portfolio, uh, dev center portfolio, then you're gonna not need to be all the way up on the CEO level and probably someone else in the C suite. Got it.

Speaker B: Ken in Growing Bay Compute, can you tell me about something that you tried that you were really confident on initially that just didn't work out as expected?

Speaker A: Yeah, we started um, by selling into the engineering side because that was something that we thought would work because they would understand the technology. Um, but what we really evolved into was uh, getting sales involved and then ultimately just moving upstream. Um, because we thought this would be one something that helps the engineer who's running a facility and it absolutely does. But the economics just weren't there for them because they weren't seeing the benefit. Aside from making their job slightly easier. Uh, what you really need is to have that economic decision, that ROI in front of the economic uh, decision maker. And that is here's the revenue that's going to be created. So uh, you know we, starting on the engineering side was something that we did initially because we thought it was something they're definitely going to understand, which they do. But at the end of the day the dollars are what matters here.

Speaker B: I always try to simplify it for people and think about ranking the buyer Personas like having a primary, a secondary, a tertiary and the one with the most part pull on that economic buyer, the one who's got the biggest vote on the decision committee goes first. Uh, the engineers it sounds like are pretty strong second. And it sounds like you even have some fans in the sales department.

Speaker A: Yeah, exactly. It's, it's been an interesting mix, um, and that tends to vary, you know, depending on the size of the organization. But uh, I think we've kind of figured uh, it out now and it's, it's typically that C FOC OO level that if we start there we can stay there. Um, the CTO is always going to have uh, an approval um, because they need to have the technology discussion and talk to the facility engineers and everything. But that's typically where we operate now.

Speaker B: Ken, when you look into your crystal ball over the next 12, 18, 24 months, what do you think is going to significantly change about what you're doing on a day to day basis with Growing Bay Compute?

Speaker A: So I think we'll see a lot more data centers start to operate in this flexibility enabled atmosphere. Uh, I think we're starting to see that with regulation, but I think more will wake up to it when they start to see the economics behind it. Um, and where we evolve to is not just the existing colocation data centers, the existing enterprise data centers, but eventually being able to serve the designers and builders of data centers. Because what we have is telemetry on all the different pieces of equipment that exist in all these facilities. And you know you have certain ah, design footprints that are top of the line spec, uh, from Nvidia example, for example, uh, their Omniverse platform shows you the perfect way to construct and run your data center. And that's your premium package and the Reality on the ground for a lot of these other sites is that they're a heterogeneous environment. And so we will have the ability to show those operators what each of those different pieces of equipment looks like, not just coming off the OEM line, but three, four, five years down the road because we have all that data. And so that's, I think where we evolve into is being able to be a platform that services more than just the brownfield data center, but also new builds, um, services, the private credit that is financing these because they want more visibility and better visibility into the assets they're underwriting. All of that data can help out. So lots of uh, different channels that can open up and different lanes we can go.

Speaker B: That's really interesting. So the data that you're collecting now, the use cases that you're building, you think is ultimately going to unlock two or three additionals in the next year or two.

Speaker A: Yes it is. And all that data stays with the customer. It's just the uh, bigger picture lessons that we are able to draw out from it. Um, but all the customers retain their own data. Everything exists on their facilities. It's just the high level summations that we are able to learn from it.

Speaker B: Final question I want to ask you Ken, is what do you think is one thing that the industry believes today that you're going to look back on in five years and everyone's going to realize was like way, way off base.

Speaker A: I think the, I think this idea that we're over building is going to be so wrong. Um, and there, there are plenty of people who don't think that we're overbuilding and think there's a lot of narratives out there that we are, um, and you know, we need to have data center moratoriums and all of this. I just think that the uh, proliferation of uh, not just large language models and small language models, but robotics and the once, once we have this proliferation across not just your early adoption adopters of tech, but your late adopters and the next generation growing up being AI native and they continue to build companies and build products. I think that there is no shortage of demand that is going to exist for the next decade. Uh, so that's one thing that I think is often talked about as AI being hype, AI being a bubble. I think we are completely underselling. You look at China, they're building 100 gigawatts of nuclear. I think we could be doing the same, but because of regulation we aren't. So I'd say that is One thing that I think is completely underestimated and missing the mark is how much demand actually exists out there.

Speaker B: So in other words, the industry as a whole is definitely not as optimistic as Wall Street. Wall Street's very optimistic. But you're thinking there may be even not as optimistic as they ought to be if they were really truly understanding all the demand that's about to be unlocked with all these additional applications beyond what we have today.

Speaker A: That's exactly right. And you see these adoption curves at rates that we've never seen before. Um, because you now have beyond human capability. Uh, you have agents now using Claude more than any human could. Right. And so you're just seeing these staggering growth numbers that I think will only continue to explode. Um, even as the cost of compute comes down, even as people uh, start to figure out more creative ways to train and run these models, you're still going to have that demand outstrip the supply.

Speaker B: Ken, this has been really cool, interesting, fascinating to hear your perspective on where the market is going. These untapped opportunities that co location providers have to open up new revenue streams. Most people at this stage look at colocation and feel like it's a mature industry and they're struggling a lot for differentiation. You're for sure giving them a way to stand out from the crowd with better utilizing the power they actually have. If someone wants to follow what you're working on at bay compute, is LinkedIn a good place to start?

Speaker A: Yeah, LinkedIn is great. Check out our website baconpute. Com but LinkedIn is great place to have a conversation. Um, connect with us and happy to go from there.

Speaker B: Cool. I'll make sure to include links to your LinkedIn profile and your company page and your website in the show notes so people can reach out to you if they want to learn more. But Ken, this has been really super fascinating. I know a lot of people at colocation operators. It's going to be a real eye opener for them to watch this. I know a lot of people that do things related to coordination alignment across the 35, 40 different sectors of data center. Tech and facilities will get a lot of value from this and even some in the construction and supply chain and investors too. This is for sure an area where they're trying to help their portfolio companies optimize for better utilizing the existing power footprint. So looking forward to this interview, going live and hearing feedback. Really appreciate you joining me today. I've been speaking today with Ken Sullivan who's the CEO of Bay Compute. Ken is based in the greater New York City area and Bay Computer goes worldwide. Thanks so much for joining me.

Speaker A: Thanks so much, Joshua.

Speaker B: Most Data Center GTM teams are flying completely blind. 83% of your buyer's journey is happening before they even speak with someone from your sales team. The Data Center Go to Market podcast is powered by DCSM M Eye. We've studied over 1900 industry leaders to build a diagnostic framework that identifies exactly why deals fall apart and revenue stalls. Stop guessing and start benchmarking. Subscribe now to our weekly Data Center GTM Briefing at www.dcsmi m.com for forward slash briefing. Again, that's D C S M M I.com forward slash briefing B R I E F I N G.