Future Forward

Seyi (00:10)
Hello, hello, hello, Reza, how are you?

Reza (00:12)
Good Seyi how are you doing?

Seyi (00:14)
I am doing well, I am doing well. Welcome to another episode of Future Forward.

Reza (00:20)
Yeah, this is going to be interesting, Seyi because we're talking about a topic that has been in the news a lot. We're going to explore data centers and AI through the lens of a city and its impact on a city and its infrastructure demands. And there's been all this news recently about AI, data centers, DeepSeq, all that. But we're going to try to do it in a way that's a little bit more broad and sort

take our usual approach of, you know, historical perspective and thinking about the future with some strategic foresight. But before we do that, let's welcome our listeners and tell them what the future forward's about.

Seyi (00:56)
Yes, yes. To our new listeners, welcome. This is Future Forward, a podcast that is really a conversation Reza and I have been having for almost 10 years now, apparently. And we discuss the past, the issues and...

all that is relevant about living and thriving in a city. We take the historical context, we grounded in the current state of things, and then we share some strategic foresight about where we think things will go. And for our new listeners, welcome. For our old listeners, thank you for coming back in.

And as usual, we hope you enjoy today's episode, which is, as Reza just mentioned, topical but timeless in our opinion.

Reza (01:49)
Yeah. So we're going to start with a little bit of historical context about data centers. And I want to use my story of what my experience of what's happened with data centers a little bit. I remember when I first went to work, servers were in closet rooms. then we had a data center at our office and it was in the first floor of our building and it was this room.

Seyi (02:07)
Yeah

Reza (02:15)
It was really cool down and all this kind of stuff. But I remember the first time that I went, like we were exploring moving our data center out of our building into a co-located facility. And I remember walking into that facility, this is about 15 years ago, and I was struck. It was this huge facility with racks and rows and stacks of servers humming away.

And the feeling that I had was, my God, this is where all the business of the internet of the world is happening. is, it was such a contrast with, I worked at a utility and I remember the first time that I went to a power plant and this was a coal power plant outside of Austin, a power project, and just being struck by the gargantuan giant nature of everything there, the huge turbine room, I went.

you know, went up the elevator, the, you know, the smokestack, saw the coal pile. That was like a different scale. And then I was in this data center, you know, a few years after that, and the scale was different. felt so, you know, so much happening in all these small spaces in a, it was, you know, was like this dystopian feeling of being in a space where you know so much was happening, but you couldn't really see it.

Seyi (03:28)
Mm.

Fascinating.

Reza (03:38)
So that was,

yeah. So all of that to say is, know, data centers have become this sort of hub of our civilizations these days where all this commerce, all this activity occurs. And, you know, to bring it to where we are today with AI and data centers and what's happening there, you know, chat GPTs launched, AI has been around for a long time, but chat GPTs launched in November, 2022 was a catalyst for

unprecedented AI demand, we suddenly felt like AI was somewhat human. We could interact with it, converse with it, and everyone thought, this is AI. Well, it's not the only, know, AI's been around, but it was, it was that moment where everyone understood or could personalize what AI was and interact with it. And it wasn't just some, you know, thing that was happening in the background. And then we've seen this unprecedented rise of stocks of like Nvidia, you know,

you know, that's the graphic card maker, but it has become the AI infrastructure kingmaker with its H100 GPU. It's really the workhorse of AI computing. Its stock has grown, you know, incredibly, I don't know, 200 and something percent. Its market cap is just under three million, even though, you know, we had that little event a couple of weeks ago. Yeah, trillion, trillion, sorry, three trillion, just behind Apple. Thank you for correcting me. These numbers are...

Seyi (04:52)
Brilliant.

Reza (04:58)
just mind boggling big. So all of this is happening and the key point that we want to talk about is that there's this change with these AI data centers, traditional data center design versus AI optimized facilities. There's some differences and it actually has an impact on this topic that we're talking about, its impact on cities, infrastructure.

You know, the power density and distribution, the needs of the power in these chips can require up to 100 times more power or up to 1,000 kilowatts per rack. To do that, because you have so much power, you need all this cooling architecture and infrastructure because these chips are generating heat, so they need to be cooled down.

And so they're using different types of cooling mechanisms, using liquid cooling instead of air cooling. It's because that extreme heat generation, immersing these servers and chips in liquids as opposed to just cooling with air. And this will be meaningful because the usage of more water over here in these cases. And also just because these

There's so much more density in these chips and servers. There's more floor loading. Floors have to be stronger. Network architecture differences and how they use space is all different. All of these are highly technical things, but all of that to say is that data centers have had to evolve with this new change in technology of AI. It is leading to some of these impacts that we're going to talk about pretty soon over here. Let me pause over here, Seyi and

you know, how do you jump in with your...

Seyi (06:34)
Yeah, no, you, thanks for providing that sort of backdrop, the background of where things have come from. And the context that I come from is on the power side of this data center surge in the U.S. and frankly across the world, but mainly in the U.S., which is where we sit. And I was reading a stat a few

weeks ago now where the Three Mile Island, is the nuclear plant, and as an aside, nuclear plants, small modular reactors or traditional nuclear has come back in fashion because of the need for power to ensure we run the data centers and the AI.

Reza (07:13)
Mm-hmm. Mm-hmm.

Seyi (07:22)
compute requirements that all these companies have. Three Mile Island, which Microsoft is bringing back up, is about 835 megawatts of power, which to bring that to terms you and I can understand, it would serve about 800,000 residents of a city. That plant would serve 800,000.

residents of a city. I worked for a power station that generated 1000 megawatts and we served roughly 600,000 residents, but it all depends on the average usage of the residents. Why am I sharing this comparison? Microsoft will be running one data center from the power being generated by this

nuclear reactor that is being brought back online. Anytime I think about that and I hear all the talk about what is required to run compute and run power to our data centers, it just blows my mind. It would be like the power station I was working for, we were only serving one specific building in the city. And that is just crazy, astonishing to me.

Reza (08:36)
Yeah, it's mind boggling. It's the feeling that I had when I went into a data center and I said, this space serves the commerce for so much, but it's in this space. And you brought that scale really well, which is I worked in a power station that served this city. And the same power station would be serving the power load for a

Seyi (08:45)
Yeah.

Reza (09:03)
data center, a single data center, powering these AI models. So it's crazy. I think that's a great way to kind of show this contrast of the world that we're living in and how transformational this technology is and how much it is actually impacting our cities. And so let me jump into that section, Seyi, and I'm gonna touch on

Seyi (09:23)
Yeah.

Reza (09:29)
a few topics like what are the resource requirements for these data centers, how these data centers are growing, what impact does it having on cities and infrastructure, and what impact does it having on urban employment. And I'll pause after each one of these sections and let you jump in. But let's start with the power consumption for these data centers, huge. Training a single large language model can use as much electricity as 100 US homes do in a year. It's staggering.

A single chat GPT query consumes approximately 2.9 watt hours of electricity, which is nearly 10 times more than the energy required for a typical Google search, which is estimated about 0.3 watt hours per query. So this energy usage varies depending on factors such as the model size, the complexity of the query, the efficiency of the hardware used to process it. But this is changing. And I think maybe let me pause here and let you jump in.

Seyi (10:25)
Yeah, so we're recording this about a week and a half after all of the US tech sector was scared into what I'll call a slightly unnecessary panic, honestly, by DeepSeek R1, which is this alternative chat GPTSK.

model that came out of China. And they have used the DeepSeek as a company, have used some approaches to require less compute for every query. So one of the approaches, I won't get wonky here, but one of the approaches is this approach called the multi-token predictions.

which is different from single-talking predictions, which, chat GPT, for example, uses. And in layman's terms, what chat GPT was doing was predicting the next word as you made a query and it was responding to it. What DeepSeq does is it makes a few guesses.

of what the words in front will be. So instead of one word, it sort of predicts the next few words that make up that sentence that you're, that it's trying to generate for you. This, and this is based on some research, this approach sort of reduces how much time and consequently how much energy that is used to generate an output. And so,

Reza (11:55)
Mm-hmm. Mm-hmm.

Seyi (11:58)
There's a lot of panic, but what it still fundamentally means as far as you and I are concerned, it doesn't change this idea that the demand for more power is there. Just the simple fact that we will have to train how we train. Yes, it matters for compute and eventual output, but we will still demand power despite

Reza (12:00)
Mm-hmm.

Yeah. Yeah.

Seyi (12:27)
whether it's DeepSeek or chatGPT. Yeah.

Reza (12:28)
Yeah.

Yeah. And to kind of put a point on that, Satya Nadella has talked about Javan's paradox, which states that when something becomes more efficient, people tend to use more of it instead of less. And so the fact that we had DeepSeq just means that it's more of a commodity, more of it will be used. So it doesn't really change the trajectory of AI.

Seyi (12:44)
Yep.

Reza (12:53)
and data centers, like we will still need data centers, you know, we'll have more efficient models. So given that, let's talk a little bit about the growth of AI data centers. You know, I mentioned that they need, you 10 to 20 times more power per square foot than traditional data centers. AI data centers are driving a surge in electrical consumption with global power demand from these facilities expected to grow at a compound annual growth rate of

nearly 48 % through 2027. By 2026, AI operations alone could account for over 40 % of data center power consumption. So if all the AI data centers could account for over 40 % of the consumption. So that's pretty interesting. Maybe these numbers will be slightly different given what's happened with DeepEak, but there still is going to be this upward trajectory. What do you think, Seyi?

Seyi (13:51)
Absolutely. Yeah. And then there's a, there's another sort of nuance there. Yes. Regardless of some of these changes and these nuance takes on some of what is going on, including this one I'm about to share. The thing we hope our audience takes from this is that there will be increased demand for power. And I was talking to some researchers toward the end of 2024 and they

made me aware and I dug into it a little bit, what Nvidia has built for the most part and has been used a lot for the most part for most of these model training and compute and inference as the case may be, are GPUs, which are graphic processing units, if I'm that correctly.

Reza (14:40)
Mmm.

Seyi (14:48)
And they were originally made for video games, honestly, but we have now modified them to better move around video, text, voice, as the case may be. The next age of processing units that will go into some of the data centers are...

AI specific processing units that are known as neural processing units, sort of like neural networks, one of the approaches to artificial intelligence. The NPUs, as they will be known, we still do not know what the power consumption of NPUs will be. We know what the consumption of GPUs will be. NVIDIA is trying to reduce

Reza (15:33)
Yeah.

Seyi (15:37)
or make the current ships they're building more energy efficient.

But we, and we've built our data centers on this current model. We genuinely just don't know what the future will hold. So it could be a lot more or a lot less, but it will be different from the current demand that was happening in the world honestly, and especially in the U.S. For about two decades, the U.S. grid did not see electricity demand growth.

Reza (15:51)
Yeah. Yeah.

Seyi (16:09)
And we've gone from low single digit growth for about 20 something years to what you're saying is about 40 something percent in a short span of time. The grid isn't really ready for this is my opinion, but that's a whole separate episode. Yeah.

Reza (16:13)
Yeah.

Yeah. Yeah.

It is a whole separate episode and let me tease a future episode where we're going to have an expert on Doug Lewin who talks about the, you we'll be talking to him about the future of the grid. So after this episode, look for the episode, you know, next week, which will, which will touch on that specific topic and, you know, what this, you know, what this age of electrification and what it means for the grid ahead of us, you know, coming back to

you know, this impact of these AI data centers. Let's talk about what we always talk about, which is the impact on cities and infrastructure. So, you know, the geographic concentration of these data centers in certain regions, they create localized stresses on electricity grids, which require upgrades in capacity and reliability. Those are obvious. We're finding these major AI clusters emerging in urban areas. You know, the Phoenix metro area has over 60 million square feet of data center space.

that is being built out. Northern Virginia, largest data center market globally. Singapore, they have put a moratorium on new data centers due to the power concerns for that small city state. Utilities face difficulties in resource planning due to the unpredictable growth of these AI-driven workloads. so utilities want to predict in the future how much energy do I need to plan for to support the grid in that area.

Seyi (17:19)
Wow.

Wow.

growth.

Yeah.

Reza (17:44)
This

last rate of growth is, know, utilities don't move fast, clearly for good reasons because they're planning 50 to 100 years in advance, but all of this is happening so quickly that it makes it really difficult for them. I got this from Bloomberg, and I'm going to quote from an article that I read that will include in the show notes, an exclusive Bloomberg analysis shows that more than three quarters of highly distorted power readings across the country

are within 50 miles of significant data center activity. While many facilities are popping up near major US cities and adding stress to an already fragile grid, this trend holds true in rural areas as well. So there's this real impact that's coming from these data centers and how much power they use with these distorted power readings. So it is starting to have some kind of impact that will have to be addressed.

And then one other impact that I want to bring up before I let you jump in is that these cooling systems for AI data centers require significantly more water. And because they need more of this water cooling, it's now competing with municipal water needs. And we've talked about this, like the nexus between electricity and water, you're seeing that over here even more so, a very, very direct tie.

Seyi (19:00)
It is. And these are areas that are near and dear to our hearts too, having worked in these spaces. I hope what our listeners take from this, thank you so much for sharing that analysis, is this message that we always try to bring to the fore when we talk about. will pick a specific topic.

But because it's part of a system in a city, the impact will be felt in multiple vectors of the system and in the city. And in some cases, the impact will be almost as serious, if not more. The water situation in the US, for example, is already dire. To add this to it is super problematic.

And we don't talk about that enough is maybe the comment I'll make here, but thanks for sharing that. does highlight the interconnectedness of all our infrastructure and our resources in cities.

Reza (20:01)
Yeah. And so one more click on what we often talk about, like what's the impact on the community? Now there going to be some local economic development impacts. You're going to have direct jobs and construction and operations of these data centers. You're going to have some indirect employment in supporting industries. And there going to be some tax revenue implications for cities because of these data centers. Obviously, we're going to see shift

in knowledge work patterns because of AI, but that's probably not a topic for this episode. That's clearly a meaty enough topic because we're gonna, job categories are gonna change, people's work and how they do work with AI is gonna change. So we're gonna put that on hold, but there are these impacts from the direct building of data centers in the communities that we live in.

Seyi (20:49)
Yeah, no, it's another really good sort of connected topic along the line of what happens in cities when one of the vectors of the system see such a massive change. The stat I picked up for this one, OpenAI, Oracle, I believe, and...

Masayoshi Sons, SoftBank, yeah, and SoftBank, they announced crazy amount of money, I think $500 billion or so for building AI data centers. And there's already work going on in one in Aberdeen, Texas, where 10 buildings are under construction. But even in the planning for that construction, they already declared that only 51 people will work.

in those 10 buildings. It's crazy. It's crazy. it will imagine building 10 buildings in a small pocket of Austin. You would expect there would be more than 50 people, multiples of that who get to work, engage and possibly utilize the space. But

Reza (21:42)
That's crazy.

Yeah. Yeah.

Yes.

Seyi (22:04)
there will be a big shift. It won't create as much human employment, even as it might increase the demand on the resources in the city, in the town of Aberdeen, yeah.

Reza (22:15)
Yeah.

That's a really interesting way that you brought that because it's something that I failed to mention when I was giving my example of the difference between going to a power plant and then going to this, you know, co-located data center. You know, at the power plant, it was full of people, like hundreds of people, you know, coordinating their efforts to make power run. And then when I went to that data center, I forgot to mention this, but like the observation that I had was there's so few people here.

Seyi (22:42)
Thank

Reza (22:42)
There's so

many people here running this thing that is running so much. Now, you know, I don't want to call judgment on, you know, hundreds of people at work versus a few people at work at data center. There's different ways to create economic value, but you can imagine that it has an impact on a city. I mean, that power plant basically sustains a couple small towns in Texas around it, but a data center is not going to do that.

Seyi (22:46)
Yeah.

you

Yep.

Reza (23:09)
I can't tell whether that's good or bad. mean, the data center is also supporting people that are doing work, valuable economic activity, but it does have an effect on employment in that city. yeah, that's an interesting thing that we're going to have to adjust. The world is going to change and this is going to be a topic for a future episode to explore.

Seyi (23:15)
knowledge workers everywhere. Yeah.

Thank

Absolutely.

We will have to exploit and it leads directly to one of our laws, which as you mentioned, the power station supported a few cities, but the data centers and there's one being built. are actually a couple being built not too far from downtown Round Rock, two data centers, but there's been honestly.

no traffic increase as a result of those data centers. But what it isn't doing is enhancing the economic benefit to the other towns and cities around Round Rock. Like Hutto, which is right next to Round Rock, Buda, which is a bit further away, not too far, Pflugerville, they are not seeing.

the benefits of that data center. And so this brought up the law for us, is that, one of our 21 laws, and this law is that regional cooperation enhances sustainability. And the idea behind this being that sustainable cities cannot exist in isolation. They must coordinate with surrounding municipalities and rural areas for truly sustainable regional development. And this is from.

Wheeler and a few other researchers, their research from 2002, where they argued for the importance of regional approaches to urban sustainability. And you gave an example earlier about the need for water in these data centers. And Brushy Creek, which I believe provides water for a couple of the data centers, is not just serving round rocks.

It's supposed to serve a few of the other cities around here. But if there's increased demand due to these data centers, that is not offset by reduction in demand in some of these other cities. Imbalances start to kick in and the sustainability of not just Round Rock itself, but the neighboring towns and cities, the lack of cooperation.

will lead to inefficient resource use, conflicting land use policies, absolutely, and the inability to address cross-boundary environmental issues that will arise as a result of the new developments that happen in these pockets of huge growth in terms of the demand for electricity and water as a result of data centers.

Reza (25:44)
Mmm.

Yeah, yeah. That really brings to point about like system, right? These data centers are causing these unprecedented changes and making a change in one part of the system is going to affect others. And so this regional cooperation is really critical to get either the benefits or reduce the harm that could come from it. So we're going to jump to this next

Seyi (26:23)
Yep.

Reza (26:27)
which sort of leads in really well from all these things that we were talking about, which is the strategic foresight and future impact. And I'm going to touch on a few topics here, like the evolution of the infrastructure and these data centers, some of the urban planning considerations, some of the policy and regulation issues that need to be addressed, as well as what impact on our communities. And I'll pause after each one and get your thoughts. So from an infrastructure perspective,

I would imagine that data center operators are going to be investing in different ways to get power than what they're doing today. You already see this with the Three Mile Island example that you shared with Microsoft, but there's going to be more than just these modular reactors. Renewable energy sources are going to be a huge part in doing this. And so there's going to be a push for more carbon neutral operations. And so

Seyi (26:59)
you

Reza (27:20)
they're gonna have to find ways of getting power and building these data centers to mitigate some of this environmental impact of expanding data center footprints. There clearly is a desire to have more data centers, but I am sure the market is gonna ask for different things to not make them as costly from a power perspective, because power is gonna be a real limiting factor.

in the ability to build out data centers, especially in the US, given some of the regulations about building out the grid and in different places and things like that. I also expect that these tech companies, these data center operators are going to find ways of reducing the energy needs. We already saw, we see that with DeepSeq. They're going to find ways to build more efficient chips. don't know. We can't imagine, but...

know, human, you know, with constraints come human ingenuity. And that's what the Chinese did with this deep seat model is they had constraints of on the types of chips that they could use. And so they found a way of building a model with less power and less money. And I expect that will continue. So, you know, even though we've seen today what's happening with data centers and AI, I expect that will be changing as we look into the future. What do think, Shady?

Seyi (28:33)
I totally agree. This is actually one of those areas where I think the AI we're building can help us improve how we build data centers for AI. Because a few days ago, I was looking at one of these infrastructure maps. And it is fascinating but unsurprising to see that there's a lot of clustering going on.

in the current build out of data centers and facilities to power the hyper scaling of AI. I looked at that infrastructure map and realized, there will have to be a lot more compute, pun intended, committed to figuring out more sustainable and optimal location planning.

that will have minimal impacts on resources, but also utilization of sustainable forms of energy generation to power the infrastructure. I speak from a personal perspective, not from my employers, because this is what we do, but it is absolutely critical for the sustainability of

the data centers and hyperscalers business model itself to ensure that as at the same time as we're drawing on this utilization of our resources, we factor in the impact on the people who live in these communities and these cities. Because if all the water and all the power is going to power in the data centers,

Reza (29:47)
Yes.

Seyi (30:09)
the rest of the community will either see higher prices for the use of what's left over. And at that point, we will start to see negative responses to what was supposed to bring economic growth, but is really causing more harm than good.

Reza (30:26)
Yeah,

and that's perfect segue into the next one, which is around urban planning and what considerations cities need to take into account. know, cities that are hosting these data centers, they need to find a way to balance the economic benefits with challenges such as what's happening with the land, you know, where's the water going for, you water consumption for the cooling these data center systems. There's going to be increased energy.

Seyi (30:34)
Yeah.

Mm-hmm.

you

Reza (30:52)
infrastructure requirements, so they might have to modernize their grids. They will have to address their water resource plan and determine where these data centers fit in with the water needs around and where they can site these data centers in places that don't impact communities in an adverse way.

Seyi (31:11)
No,

totally. It's very well sort of framed. If you listen to the urban planning episode we had, it's stuff like this. A lot more consideration for the market forces in planning our cities. So if you didn't listen, you should listen to that because it ties to the point Reza just made.

Reza (31:29)
Yeah.

And then the other one is closely tied to it, which is around the policy and regulation. Governments are incentivizing or need to, even more so, incentivize sustainable practices around the build-outs of these data centers. Many want sovereign AI capabilities and so want to incentivize building those data centers in their countries or their areas so that they can maintain competitiveness while

while at the same time they need to do this in a way that addresses environmental concerns. I also believe that there needs to be more local governance around these AI data centers because of the impact on communities. And we need to upgrade our environmental impact assessment requirements to address this new need. I'm going to click into the next one too before I let you jump in.

And the reason it's because we need that type of policy and regulation because of the impact on our communities. Location selection can be an issue. It can cause tension between the power availability and the workforce access and where they're citing these. There could be environmental justice concerns around where those facilities are cited. It could be the thing I talked about, the power disruption caused by data centers close to those communities. They get cited in places that are

less advantaged and causing more adverse impact over there. So there could be community impacts, but at the same time, cities and their communities need to think about what can we be doing to gain benefit from this? Can we be building better green energy infrastructure to power these data centers to where our community is at the forefront of sustainable

sustainable build out of these data centers and that pushes forward sustainability in the city itself. What do you think Seyi?

Seyi (33:10)
I absolutely agree and I'll sort of summarize based on a conversation I was having with a gentleman a few weeks ago and we're talking about development of, in this case it was some solar stuff again, this is outside of the job I do every day and he goes, what many of these companies will realize very quickly is that

If the community doesn't want it, nothing will happen. I was like, wow, that's you apportioning a lot of power to the people in a community. goes, yeah, that he's realizing people actually have a lot more say in whether some random data center gets plunked in their neighborhood or not. And

that the accountability that we seem to not see in a lot more of the decisions that are made at the federal level or at the state level, that isn't the case when you come to a city and the guy who's the deputy mayor of the city grew up in the city and his family members are scattered around the city and consequently his decisions.

will impact his community.

Reza (34:33)
Yeah,

yeah, that's really good. It's so interesting, Seyi, I think, you I was trying to think of like, what's the best way to have a call to action like we do at the end of every episode. And this one, you know, my call to action is not necessarily about sort of data centers, but it's more around AI. I'm curious to hear from our listeners about how are you using AI in your daily life or your daily work?

Seyi (34:59)
Yeah.

Reza (34:59)
Because

I want that to kind of inform this topic that we haven't touched on in this episode is like how is AI actually going to change our work or our lives? And so I you know the call to action is to share that with us, you know, we have our newsletter or email hello at futureforward.fm So if you could share back what you're experiencing with AI And maybe we could sort of pull on that thread and have an episode based on that

Seyi (35:08)
Yeah.

Absolutely, I really like that. And on that note, I can't even think of a better way to say, I hope you've all enjoyed this episode. Please reach out and share. It is a topic that will be with us for a while, I can imagine. And so your perspectives, if you have expertise on this.

Please reach out, give us your thoughts and we might just have you on to dive deeper into the topic if you will. thank you for listening and Reza, thanks for all the research on this one. It's a fascinating topic.

Reza (36:01)
It is, it is. So Seyi, do we have any mail bags for this episode?

Seyi (36:06)
No mail bags from me, yeah.

Reza (36:07)
Okay,

so, you know, we'll do what we do at the end of every episode, which is please take a minute to rate and review our podcast. That really helps people find it on the podcasting apps. Take a minute to like and subscribe. We now have over 6,000 subscribers, Our minds are blown. We took a look it earlier this week and I can't believe it that we have a lot of

Seyi (36:29)
Blown. Blown, yeah. Yeah.

Reza (36:34)
people, so many people.

So become one of those 6,300 people that are subscribed to our YouTube feed. And then most of all, the best way that we grow is if you take a minute to share this with someone that you think would find this interesting, someone that you care about, that's always word of mouth makes a big difference. And with that, Seyi, you know,

We come to the end of another fantastic episode, at least we think so. We have so much fun doing it. I hope everyone enjoyed it and we look forward to seeing you on the next one.

Seyi (37:02)
we do.

Yes, bye listeners, thanks again.

Reza (37:10)
Thank you.