Mahdi Yahya (Ori / Radiant) on Getting Acquired by Brookfield, Building the Backbone of Sovereign AI, and Why Intelligence Is Infrastructure

# FWDstart Podcast: Mahdi Yahya - ** COLD OPEN** **Mahdi:** Intelligence in general is becoming infrastructure. When you build the electricity grid of a country, that's a sovereign asset that you need to control. The cost of over-investing is going to be cheaper than the cost of under-investing. Theatre and film, especially drama in general, is a great way to explore humans and I think it helps you sell later on in business.

We don't care how many hours you work on something, as long as you produce exceptional work. You can work for an hour a week. Experience does not mean exceptional work. It was 2006 where the airport was bombed.

The closest nearest airport was Damascus. So I took a taxi. Just stopped the cab. He agreed to take me to Damascus.

When you're raising money, you only need one door to open. You knock on a thousand and if one opens, that's all you need. - ** WHAT IS RADIANT AND THE BROOKFIELD DEAL** **Jamie:** Ori has been acquired by Brookfield Asset Management. Congratulations.

**Mahdi:** Thank you. **Jamie:** On that enormous achievement, and has been merged into Radiant, which is a vertically integrated AI company. How would you characterise it exactly? **Mahdi:** We've been going as Ori for about eight years.

We focused on building the best software in the world to run applications across high-performance computing infrastructure, distributed infrastructure and so on. Over the years we built the best team in the world to deploy and manage and run large-scale AI infrastructure around the world in a number of data centres that we operate. Very quickly you realise, especially after the AI boom, that to scale this you require vast amounts of capital. You require some creative ways of deploying this capital as well to build more infrastructure.

And you also require vast amounts of energy and power across data centres. ** ** Brookfield is one of the largest asset managers in the world. They're one of the best capital allocators and deployers in the world. They also own one of the largest power land banks in the world.

So that's really where the rationale came for the deal and the conversations we started having with the Brookfield folks. If we combine these four elements together, we can bring to market a new, strong value proposition to the AI infrastructure build-out. And that was the birth of Radiant, basically. Radiant combines power, land, capital, compute capabilities and software altogether in one value proposition.

**Jamie:** Was there any temptation to just continue on your own trajectory? You raised a pretty significant round around this time last year. It was 140 million, I believe. **Mahdi:** Yeah.

Even that's not enough to scale. The space we're in, throughout the process we had offers from multiple other investors when we were talking to Brookfield as well, offers in the hundreds of millions plus. And I wouldn't say that under the Radiant banner we're not moving forward on our own trajectory. It's still the same team, same ambition, now with bigger partners to ** ** deliver on that ambition.

- ** IS THE FOUNDER'S JOURNEY OVER?** **Jamie:** Is the founder's journey over now or is it just beginning as far as you see it? **Mahdi:** No, it's just another phase in that journey. There's many ways to define a journey.

There's the boilerplate way that everyone does today, raising multiple rounds with marked-up valuations with different VCs and so on. Yes, it could work for a specific type of industry, but the space we're in, what we are doing, and not just ourselves but our peers in the industry as well, has never been done before. Building infrastructure at this unprecedented scale and combining traditional real estate investment with technology investors and VCs and everything that sits in the middle, that's the space we're in.

** ** So going through a traditional founder journey of how you finance this and get to an exit or an IPO does not apply here. - ** MAHDI'S ARTISTIC BACKGROUND** **Jamie:** One tension I was keen to explore is that during the course of research, you've got a super unique background as it pertains to the artistic side of things. You studied at the Drama Centre in London, which I'm looking forward to asking you more questions about. I was listening to another podcast you did recently where you compared your work with Ori to artwork, in the vein of Dalí, Picasso.

I do want to touch on branding, design, and aesthetics later as well, because Ori in particular did something really interesting as far as that's concerned. I think it was after the Wa'ed Ventures deal, sometime last year, I started to get ads on LinkedIn for Ori, and I was like, this is a really unique blend of influences. But we'll get to that in a bit. I think it was maybe Rashed from Brookfield who called Radiant something to the effect of a leasing business with no technology risk.

** ** Do you agree with that description and framing? Does that not maybe jar with the artistic characterisation? **Mahdi:** No. Building large-scale infrastructure and deploying that in data centres is a form of a leasing business.

And that's what we do. We are the best in the world at doing that today. In order to deliver a good leasing service to your customers, there's technology embedded into that, there's software, there's services, and so many things. But on the surface, you build solid foundational infrastructure for AI, that's what we do, and we lease that to customers and partners.

So I do not disagree with that. Now, merging that with the artistic side, my view is always let's add some character to things. That's how I look at it. Cloud infrastructure in general is, I mean, I find it a very exciting business, but in general it's ** ** a boring business for the majority of people, because it's invisible.

And it powers everything we do today. It powers the internet. Without it there would be no internet, there would be no Instagram. Everything.

**Jamie:** We couldn't doom-scroll into oblivion. **Mahdi:** Exactly. There'd be no podcast. **Jamie:** This is true.

**Mahdi:** But it is a boring foundation for the internet, and sometimes it's nice to add some character to that, within reason. That's what I try to do when we come to how we design things and how we present ourselves. It's not always easy. Sometimes it doesn't work.

Sometimes it works. But it's just a fun thing to do, I feel. **Jamie:** I couldn't agree more. And I know you've got a passion for film.

Funnily enough, I'm a film studies graduate. **Mahdi:** Oh really? **Jamie:** Yes. So I think you mentioned Tarkovsky and I saw some parallels between the branding, ** ** it kind of reminded me of Solaris, in terms of what you're trying to accomplish.

It's like nearly marine biology-inflected. I don't know exactly how to characterise it, but that was the point. **Mahdi:** Yeah. The name Ori itself comes from biology.

It's the origin of replication of DNA. In science, they use ORI, and that's where the name came from. It's kind of like the permission for biological creatures to start replicating themselves. That's how we saw it.

Like a genetic code that connects all living beings. The parallel, when we were thinking about the name, was: what's the genetic code for infrastructure? Can we build the genetic code for infrastructure? And that was the idea behind the name.

- ** WHAT BROOKFIELD UNLOCKS** **Jamie:** What does Brookfield unlock for you meaningfully? Does this massively accelerate the roadmap and what you're able to do? Or are you continuing along the same track? **Mahdi:** The ambition is still the same and the plan is still the same, to build large-scale infrastructure around the world.

Because we believe that, well, not even believe, we see it today, we experience it: the demand for AI is not slowing. **Jamie:** Insatiable, I'd imagine. **Mahdi:** Constantly growing. It does compress timelines.

It gives us the ability to move quicker and the ability to deploy bigger as well. **Jamie:** What does it actually look like practically speaking in terms of geographic distribution? **Mahdi:** We're focusing on three geographies today: Europe, given that we are headquartered in London; North America; and the Middle East, where we've been working for a while with a number of partners. - ** LEAVING LEBANON AT 19: THE TAXI TO DAMASCUS** **Jamie:** Let's rewind then, because you have a connection to the Middle East as well, originally Lebanese.

Can you tell me the story of how you came to London? I've seen a short version: you leave at 19, 2006, taxi to Damascus, fly to London. Can you expand on that story? Because I feel like that doesn't adequately do it justice.

**Mahdi:** I mean, it's as simple as that. **Jamie:** No, come on. **Mahdi:** It was the 2006 war in Lebanon. I had ambitions to start a company or start something.

**Jamie:** Where did that come from at that age? **Mahdi:** My dad instilled that in me. He was very entrepreneurial himself. A lot of Lebanese are entrepreneurial, by the way.

His father before him was, in his own way, also entrepreneurial. So it's kind of instilled. It was the 2006 war. The airport was bombed, so I wanted to ** ** leave, but the airport was bombed.

The closest airport was Damascus. Back then Damascus was very quiet, none of what we've seen over the years. A lot of Lebanese actually fled to Damascus for safety, or to Syria in general. So I took a taxi.

Just stopped the cab and he agreed to take me to Damascus. **Jamie:** How long a journey is that on a good day? **Mahdi:** Probably around two hours. I just found a ticket.

I think the ticket was not to London directly, it was to Vienna first. Then you had to stay in Vienna for a bit, and then another ticket to London. **Jamie:** Was the ambition always London? **Mahdi:** I've been to London before many times when I was younger.

I actually visited my first data centre ever here in London. I think it was when I was 14 or 15. It was easy for me to get here, I speak the language. That helps.

I've always loved London. - ** ARRIVING IN LONDON AND SAM TELECOM** **Jamie:** So you arrived in London. You don't have a degree, you're 19. Sam Telecom.

Talk to me about it. How does that happen? First of all, what's the interest in data centres? **Mahdi:** My dad worked in telecoms.

So my visit to the data centre in London was through a trip. **Jamie:** That makes more sense. I thought you just wandered into a data centre. **Mahdi:** I've always been interested in tech since I was a kid.

I had the first computer in school and all the school kids used to come to my house to see the computer. So I just grew up in that ecosystem. That helped. Sam was the first company.

It went through so many iterations, but the original thesis was always the same: looking at networking and moving traffic between data centres and networks. ** ** Building on the experience I'd learned through the environment I grew up in. I was lucky because, growing in that environment, you can lean on people, put things together and start a company. So I count myself lucky to be able to do that.

- ** WHY DRAMA SCHOOL?** **Jamie:** Let's talk about drama school then. Why on earth do that? **Mahdi:** I've always had two interests in life: the arts and technology.

And I don't see them as different. There's a lot of overlap in so many ways. When I came to London, I worked first and did my first company, and then I wanted to go to uni. I thought, I'm not going to learn much at business school, and if I go do a computer science degree, I don't have the patience to go through the whole programme.

** ** I'm never going to be a coder or an engineer per se. I understand engineering and engineering concepts, but I wasn't going to practice as one. I was in London, which is the theatre capital of the world. So I decided to go to drama school and went to Drama Centre London.

But also, it was a way to explore humans. Theatre and film, especially drama in general, is a great way to explore humans, and I think it helps you sell later on in business. **Jamie:** I entirely agree. I went to film school.

Russian cinema, post-revolutionary Iranian cinema, French poetic realist cinema. That was my thesis. Marcel Carné. And again, to your point, I found it extraordinarily complementary.

** ** I think even just purely engaging with people and pitching, selling, funnily enough. I think it's enormously helpful and vastly underestimated. Having the space for four years to pursue one's passion was extraordinarily useful. Original ways of thinking about things.

I wouldn't change a thing. **Mahdi:** Me neither. Understanding how to tell stories is a very powerful tool. I think every stellar 10x founder is an amazing storyteller.

There's no coincidence. The way I look at storytelling, you're not just telling stories in words. You tell stories in so many different forms. When you're building something, in the design you put together, what's the story you're telling there?

What you're writing, the teams you put in place, there are all these different elements of storytelling that you use while you're ** ** building a company. - ** ROOM ONE: ART MEETS TECH** **Jamie:** How does Ori come about then? Fill in the gap between drama school and Ori. What's that transition?

**Mahdi:** Before Ori, there was a company called Room One here in London. Room One was a financial disaster, but it was the most fun I've had. **Jamie:** You're going to have to expand on that. **Mahdi:** It was my idea of, hey, let's try and mix the arts with technology together and see what happens.

That's why it was a lot of fun. I met a lot of fantastic people through that, got involved in some fantastic projects. We had a gallery space, studios underneath, residencies and all the kinds of things you'd expect from a place like this. It was an experimental ** ** phase in my life and an experimental place for lots of people.

At one point we started working with Ericsson, working with technology companies and the arts to see how we could apply that. What we did was we started building with Ericsson what we called 5G use cases. This was before 5G was robust. The idea was to demonstrate what 5G can do, and we decided to try and demonstrate that through art.

We built virtual reality experiences. 5G is fast and super low-latency, so virtual reality can work better and smoother. We built remote surgery experiences as well. This went to events and different places.

This partnership with Ericsson was one of the many different things we had, but I'm focusing on this because this is what led to Ori. ** ** This is where I started thinking that compute in the cloud might get distributed. If networks are becoming faster, then you can move more data through them. There are different types of packets and applications that we can leverage.

And therefore maybe we need to build the genetic code of infrastructure, going back to the previous point. That led me to explore more of what edge computing is. There was a whole movement back then around edge computing, tied to 5G to some extent: moving processing power and compute to the edges of the network so we don't have to send it all back to centralised data centres. Whether you're running applications at the edge or in a data centre, you still need that genetic code to run applications in a distributed manner.

**Jamie:** Was this very much focused on the telco side of things? **Mahdi:** There was a large element of telco and other forms of edge. But the edge of the network is the ** ** edge of the telco network. So that's how the original foray into Ori started, and we did some tests and thinking and eventually launched it as a separate company.

**Jamie:** What happens to Room One then? **Mahdi:** Room One is tucked in the memory as a phase. It was a lot of fun. But commercially there was no way to scale that.

Whenever art is involved, invariably it's going to be constrained. - ** RAISING MONEY FOR SPECULATIVE TECH IN THE UK** **Jamie:** Was this a tough sell at that particular moment in time? What year are we talking, 2018? Was this idea of edge computing in vogue then?

**Mahdi:** This was a hard sell. We went to so many investors back then. We were lucky. One of them, Episode One Ventures, a UK fund, they're great friends now, been partners for a very long time.

** ** They saw the vision and said, okay, let's see where this goes. When you're raising money, you only need one door to open. You knock on a thousand and if one opens, that's all you need. **Jamie:** What were the 999 others saying?

**Mahdi:** A number of things. Especially back then, the UK is not known for building speculative technologies, or deep tech as they used to call it. There are a lot of spin-outs that come out of the best universities here that are quite deep tech in their own form. Arm, Graphcore, ** ** all of these companies came out of here.

But it was hard to build, especially in the UK back then, on speculative technology. There are two ways to build a company. You see a very pressing problem right now and you build a product that solves that problem. That's how you make money.

Or you imagine a problem that might happen in the future, and then you build a technology that may or may not solve this problem. You're building a technology looking for a problem to solve. We were in that camp. And that camp is very hard to raise money for, especially in Europe back then.

**Jamie:** Venture capitalists, a bit of a misnomer in terms of not being overly adventurous. And to your point, I think it is changing. **Mahdi:** AI changed that as well to some extent. But we're still not as advanced as in the US in terms of putting investments behind these technologies.

Historically, tech in Europe was very services-driven. ** ** Websites, delivery apps, marketplaces. There's nothing wrong with those, they're great businesses, great companies. They solve problems in the market and they help a lot of people.

Some of them grew to become successfully large companies. But that's the type of tech that a lot of people here appreciated and understood. - ** THE THESIS EVOLVES** **Jamie:** I suppose that original thesis has evolved over time. Would that be correct?

**Mahdi:** The technology we built or started building in 2018 is still the foundation of what we have today. Back then, machine learning was one of the use cases we were looking at. ** ** The simplest example everyone used was: if there's a self-driving car on the street, you need to process that data as close to that car as possible because adding latency to that decision-making could kill someone. **Jamie:** Not ideal.

**Mahdi:** It was a simplistic example, but it helped the layman understand. So this was one of the use cases for machine learning at the edge. That took us into AI and we started exploring what AI applications require to run on, like GPUs and so on. But it was one of the use cases we were exploring.

When you're building a technology and evolving with the markets, you're thinking about which use case is going to take off first so you can focus on that one and build around it. We were exploring multiple use cases, and that was one of them. When AI started to grow and grow, we realised that this is the use case we've been building for. - ** THE CHATGPT MOMENT** **Jamie:** When did you really start to feel that?

Was this pre-ChatGPT? Where in the timeline? **Mahdi:** It was pre-ChatGPT, probably by a year, but we hadn't really gone all in on it. We were exploring, working with a number of AI companies around running their applications.

Then when ChatGPT became the defining moment for AI, like the Big Bang, we started to put more and more effort and money behind it. **Jamie:** Did that change things from a capital needs perspective? You recognise, okay, AI is the use case, this is what we're going to go all in on. Do you need cash to make that vision a reality?

**Mahdi:** Of course. What changed for us is that at one point we realised the software ** ** alone that we'd built is not going to be enough to scale. We need to build our own infrastructure underneath it. And the software gave us a competitive advantage over anyone else entering this space, because we'd been building it for years.

We have a very deep understanding of the technical stack within the software, but also the technical stack within the physical infrastructure that AI is running on. Once you do that, it changes, because CapEx becomes heavy. You're buying Nvidia GPUs and they're not the cheapest thing out there. That changed the capital needs of the business.

It changed the type of investors as well. But we were lucky that we had some really supportive investors on the cap table already, and they helped us through that transition. We raised some money from existing investors to begin with to prove it, and then we raised subsequently from ** ** external investors. And the nice thing about when you start dealing with CapEx is that you can raise debt as well.

You don't always necessarily need to raise equity. We raised some debt against these assets too. - ** BUILDING THE TEAM** **Jamie:** One thing I'm realising we've glossed over is building out the team for Ori. Can you talk to me about that?

First hires, what you're prioritising, where you're sourcing talent. **Mahdi:** Building out a team is probably one of the hardest things you can do. Finding the right people, and the right people sometimes don't necessarily need to be the best in their field. They need to be the right people for the team, the company, the phase, the stage you're in.

**Jamie:** It can be a particular moment in time. **Mahdi:** Exactly. We were lucky that we managed to attract some really good people. We have people who've been with us from day one and are still with us, now part of Radiant as well.

** ** When you're building something, there's a pre-becoming-a-company and post-becoming-a-company phase. What I mean is, it's a misconception a lot of people have: when you start something, you raise some money, you think you're a company now. And I don't think you are. That means you don't need to hire the best HR person and the best head of people and the best accounting person, all these functions that make a company.

That actually creates less of an attractive environment to bring in the right people at that time. When you're starting out, especially when you're trying to figure out where this technology you're building is going to land, you need creatives in their own fields. ** ** People who can do things, who can think, who can work within small groups. **Jamie:** Less restrictive.

**Mahdi:** Exactly. A team changes over time. You have people who can evolve with the company as it evolves, and you have people who cannot. They're right for the business for a specific period of time, but then they're not the right people moving forward.

We were quite lucky that the majority of people who joined us managed to evolve with the business, and that's a result of creating the right environment for them. I always see my role as creating the right environment for these people to work together in the right way, and just minimising the chaos as much as possible. There's always chaos, but try to minimise it. **Jamie:** Directing the actors.

**Mahdi:** Exactly. - ** CULTURE AND PRINCIPLES** **Jamie:** How would you characterise the culture? **Mahdi:** Culture at a company, or between a group of people, in my view, is how this group makes decisions around their people. How do you hire?

How do you fire? What does success mean for that specific group, not the individual, the group? And what do we accept and what do we not accept as a group? That's what culture is.

You can go on a website and write some really nice-sounding words, and now you can use AI to write these for you, but that's not really culture. This is something we distilled early on. We don't have values, we have principles. These are principles specifically based on this criteria.

We're very direct. We like directness, we like openness. If things don't work, we say they don't work. We do not make decisions through consensus at all.

There's one person responsible for making decisions for that specific ** ** topic or subject. Whoever is the most knowledgeable about that specific topic is the one who's responsible for making that decision. Makes us move quicker. - ** ON FIRING** **Jamie:** Something you mentioned there was firing.

How quickly do you make that decision when you recognise it? **Mahdi:** Sometimes not quickly enough. **Jamie:** Do you have any lessons reflecting on previous experiences? **Mahdi:** Right now, my view is that if somebody's a maybe, then you part ways.

** ** In a humane, reasonable way. Not just dropping people like in films. You decide to part ways. Sometimes it's not their fault.

Sometimes the timing, the type of people within the group already, what the business needs, they just don't align. So it's not about the person's inability to do something, it's just the circumstances did not align. - ** HIRING FOR EXCEPTIONAL WORK** **Jamie:** And the flip side, hiring. It's difficult, as you said.

A lot of the time it's maybe intangible things and it's difficult sometimes to explain what it is that works in someone. **Mahdi:** It is. It's hard to hire exceptional people. Experience does not mean exceptional work.

One of our principles is that we don't care how many hours you work on something, as long as you produce exceptional work. ** ** You can work for an hour a week, and as long as results are there. Results over rules and hierarchy. So to find someone who can produce exceptional work is hard.

No matter how many logos they have behind them and how much experience, that's not normally an indicator of the ability to produce exceptional work. But one common trait I've seen is around the attitude. You can have people who maybe haven't produced exceptional work in their life yet, but they might have the right attitude within the right environment to do that. Over time I started to look for that more and more within people.

Looking at the data, who was able to stay with us and be successful, and what was common between them. It always went back to the attitude piece. **Jamie:** Do you think even the definition of exceptional work will evolve? We've had this idea of the one-person unicorn.

The amount of work that one individual can now get through ** ** versus one year ago has scaled by a colossal factor, and one would assume this is only going to continue. **Mahdi:** The definition of exceptional work could mean so many things, it depends on the environment. Are you making the right decisions all the time, or the majority of the time? If you're making a series of bad decisions, that's not exceptional work because you're not moving the group forward.

Attention to detail is extremely important. It goes back to storytelling. Exceptional work is the ability to explain complex matters in a simple way, depending on the audience. Details within that matter so much.

** ** And then the adaptability of being able to listen, to the words being said to you, to the environment you're in, and being able to adapt with that and constantly grow. These are not easy characters to find. And then there's the attitude: you're not phased by problems, because there are always going to be problems, and they're going to grow and grow. It's not a problem.

It's just a hurdle you're trying to get over. **Jamie:** Glass half full versus glass half empty. **Mahdi:** Whether you're a one-person company using all the AI tools in the world, or you're part of a big team, you'll still have the same issues and you still need these qualities to produce exceptional work. **Jamie:** How big is the team actually?

**Mahdi:** We're about 60 people at the moment. ** ** Not a very large team. Pretty lean. **Jamie:** I was going to say.

**Mahdi:** We're hiring a lot at this stage, but we try to keep it lean. Sometimes you don't need hundreds of people if the business doesn't need them. - ** AI AS A PRODUCTIVITY TOOL** **Jamie:** Has AI helped from a productivity perspective? **Mahdi:** Of course.

AI is great at getting you 60% there. But you still have to put in the work for the other 40%. And I don't know if we'll get to a time where it can do the 100% in a way that constitutes exceptional work. If somebody in our team is just using AI, copy-pasting what's there and putting it out there, that's not enough.

You still need to take what AI has given you and put it in the right context ** ** and in the right way within the wider context of what we're trying to achieve as a team. That's not there yet with AI. **Jamie:** That's one of the big things levelled against it, isn't it? Where are the new discoveries as it relates to science?

Where's the genuine new insight? **Mahdi:** We haven't seen that yet. But we will. And I think it will still require human creativity.

The best scientists are very creative because they have to imagine the world around them, the possibilities, and then put it into a test and examine it. I don't think AI can imagine the world around it in the way we can. Same goes to the art world. Yes, AI now can produce text like Shakespeare, but it is like Shakespeare.

It's not a new thing. ** ** Individuality in general is changing. It levelled the playing field for a lot of people. But then how do you create individuality next?

The best scientists are very unique individuals. The best artists are very unique individuals with specific traits, able to take what's out there in the world and put it together into something new. Even the best entrepreneurs and CEOs, they're unique individuals. You take the ingredients in front of you and you put them together into something new.

AI is another form of these ingredients. It will accelerate things to some extent, but you still need to put things together. I don't feel like that's going to disappear anytime soon. So defining individuality is going to change, and that's where new scientific exploration will come in, for example.

**Jamie:** So there's still a place for us. **Mahdi:** There will always be a place for us. - ** SOVEREIGN AI AND THE GULF** **Jamie:** Let's talk about sovereignty then, and maybe we can touch on the Gulf via that as well. We've obviously seen a significant push for data centre build-out across the region.

In Saudi with EXAT and Rook, in the UAE obviously with Stargate. It's been characterised as a sort of triple advantage: cheap power, geographic centrality, and sovereign capital. I'm curious, in the context of everything happening at the moment, does the current conflict alter the calculus? **Mahdi:** Look, the AI adoption movement around the world has got a lot of countries, sovereigns, even large companies and organisations to rethink their data strategy.

** ** Where does their data reside? Where is it being processed? Who has access to it? Who doesn't?

That's natural, because AI is all about data. That's the beginning of the concept of sovereign AI, and I don't think that's going away. It's going to continue to grow. A lot of countries around the world are now thinking about the sovereignty of their AI.

The most valuable part of the data isn't the existing data you're feeding in. It's the data that AI will generate over time through interactions. And that's a very valuable data point. That needs to be controlled and protected, because it touches all the way down to national security interests.

We have a solution for sovereign AI. We're working with a number of ** ** governments and organisations around the world today around what sovereign AI could mean for them, and how we can help. **Jamie:** One of the things we've covered quite extensively over the last number of months is the push to create sovereign models. Do you think every country should have their own sovereign model?

**Mahdi:** I don't. I mean, this goes down to the definition of what sovereignty means, and we could talk for hours about that. ** ** There are so many different opinions about it. There are many ways you can achieve sovereignty.

And the question should be: why do you want sovereign AI in the first place? Then you work backwards from there. If it's around protecting your data, for example, the simplest example, there are many ways to achieve that without having to build your entire stack. If you want to go with building sovereign models, that means you have to go build your own sovereign silicon and everything.

The value chain between the model and everything else is a million parts. **Jamie:** Invariably you're going to be reliant on external players. The value chain just isn't there to create a vertically integrated company in a given country. **Mahdi:** Even the largest US companies, not everything is built in the US.

But it's about who controls it and how do you put the right frameworks in place ** ** to control that stack you're building. If you can't answer why you want sovereign AI, then you shouldn't really be looking at it. **Jamie:** When government entities or companies come to you interested in the sovereign side, what's their answer to that question, more often than not? **Mahdi:** It's around the data.

It's around controlling the stack. Because intelligence in general is becoming infrastructure. When you build the electricity grid of a country, that's a sovereign asset you need to control. If it's in another country and gets switched off, you have no control.

The water distribution system is a sovereign asset. The telecom networks are sovereign assets. And intelligence now is a sovereign asset as well. Defence runs on that.

National security runs on that. Healthcare runs on this ** ** infrastructure. If you're going to be running national security, healthcare, the education system, the financial system of your country on this asset, then you need to think about AI as a sovereign asset, same as electricity and water and everything else. Then you need to control that stack.

You don't need to build that stack all yourself, because when you build an electricity grid, you don't build every single component. **Jamie:** It would be prohibitively difficult. **Mahdi:** It's the same thing. It's about controlling the stack.

That's the second layer of infrastructure assets that any country can have. Intelligence is infrastructure today. **Jamie:** I think that's a really useful framing, even as it relates to sovereignty. You've characterised that really well.

- ** WA'ED VENTURES AND THE MIDDLE EAST** **Jamie:** Can you talk to me about Wa'ed Ventures? That was the first time Ori came onto my radar, because I had covered their investments. I think it was over a year ago at this stage. How does that come about?

**Mahdi:** I think one of our investors knew someone there and I got introduced. Obviously I'm from the Middle East and Saudi was a market we were seriously looking ** ** into. I think they're doing fantastic things in Saudi and the Middle East in terms of how the economy and tech scene is growing. I wanted to be part of that.

So I met with the Wa'ed guys, and they're great investors, extremely supportive since day one. The idea was that we're setting up operations in Saudi as well. They came on board as an investor in the parent company here in the UK. **Jamie:** Will that continue under the banner of Radiant?

**Mahdi:** The Middle East is a big focus for Radiant strategically. I was there about a month ago. We went to Riyadh, Abu Dhabi, Dubai, Qatar, different places. - ** CROWNING ACHIEVEMENT OR SOMETHING MORE?

** **Jamie:** Do you think Ori and Radiant, is this your opus? Your crowning achievement? Or is there another problem you'd love to tackle further down the line? **Mahdi:** There are always problems to tackle.

But you would only know what your crowning achievement is when you die. That's the unfortunate reality. **Jamie:** You wouldn't know. **Mahdi:** No.

Maybe by then there'll be AI that you can communicate with. Some sort of setting you'd have been plugged into. **Jamie:** Are there any other problems that you're particularly interested in? **Mahdi:** Building AI infrastructure over the last few years has opened my eyes a lot to how underdeveloped the energy sector is.

** ** I've had a lot of ideas there in terms of how that can evolve, especially with the rise of AI infrastructure that is hungry for power and energy. There's so much that can be done in that space. **Jamie:** Are we talking renewables? **Mahdi:** Renewables is just the Formula One.

I'm talking more about how we deliver energy, how we distribute it, how we connect it, how we consume it, how we make it more efficient. - ** ARE WE OVERINVESTING IN AI INFRASTRUCTURE?** **Jamie:** Are we overinvesting in AI infrastructure? **Mahdi:** Somebody said once that the cost of over-investing is going to be cheaper than the cost of under-investing.

And I think that's correct. I don't think we are overinvesting. We have to believe that the AI adoption curves are only going to increase. And I fully believe in that.

It's not going to slow down and it's not going to go away. The genie is out of the bottle, and ** ** it's going to be pretty difficult to put it back in. It's evolving and getting better. We're dependent on it entirely in so many ways.

When the cloud goes down, what are we supposed to do? That's a bad thing. But you have to believe that the infrastructure is going to scale from here. I don't think we're overinvesting.

I think we're under-investing. **Jamie:** I don't think anyone wants to be the pessimist either. And I'm sure there will be companies that are burnt and it doesn't quite work out. **Mahdi:** But that doesn't mean the overarching thesis is wrong.

We'll get things wrong so many times as a human species in terms of adopting this. But it's not going to go away. This is the next development of us as a civilisation. - ** BUILDING FROM LONDON** **Jamie:** London and the UK maybe sometimes gets a bad rap, maybe unnecessarily.

I disagree with the characterisation, the idea that it's a living museum and its best days are behind it. Can you talk to me about building the company from London? **Mahdi:** London is greatly positioned from an international perspective. It's a metropolitan city.

There are people from all over the world. Everything happens here. You can find anything here. The quality of life in London is fantastic.

A lot of people have been saying recently that London is degrading, and cities go through ups and downs, but the quality of life is still much better than so many parts of the world. ** ** That's an important environment to build something from: access to an international environment and a good quality of life. From a time zone perspective, it's fantastic because you can work with the US, you can work with anywhere in the world. And there's a huge cluster of really good talent here.

It's a magnet. It's been great building a company from the UK. I owe the UK a lot for what I've done over the last years of my life. Being here allowed me to build what I've built so far.

**Jamie:** Do you think it would have been easier to build Ori from San Francisco? **Mahdi:** I don't think about these things this way. I just think, this is the journey. Would it have been easier?

Maybe we'd have raised more capital. But I don't know. Maybe we'd have raised ** ** so much capital that it became extremely bloated and we wouldn't have even survived. Sometimes hardships become positives in the future.

**Jamie:** This is very true. Sometimes when you raise a lot of capital, it makes it very hard to change and evolve. COVID is case in point as far as inflated valuations. I think we're seeing that play out in a regional context.

We have a lot of startups who raised at extravagant valuations, enormous amounts of capital from international investors, and they've been very quiet the last three years because it's just not possible to live up to that. **Mahdi:** San Francisco is a great city. The biggest tech companies come from there. Clearly there's a great environment to build a tech company there.

But I didn't build a company from there. And that's how I look at it. Everyone has their own journey. Everything happens eventually for a reason.

- ** HIRING AT RADIANT** **Jamie:** You're hiring quite a bit at the moment. How aggressive is the headcount scaling? **Mahdi:** We're hiring constantly. **Jamie:** What roles have been prioritised?

**Mahdi:** Engineering. A variety of different types: data centre engineering, infrastructure engineering, software engineering, product engineering. We're hiring a bit of marketing and some supporting functions. But primarily engineering is where we're scaling.

We're an engineering company. - ** THE SALES CYCLE** **Jamie:** What does a sales cycle look like? I come from a sales background as well, so I'm curious. Is this a really protracted sales motion?

Who's the ICP? Unpack that for me because it's one of those vaguely impenetrable things from the outside. **Mahdi:** In AI infrastructure specifically, there is no set sales cycle. ** ** I closed a $40 million deal in a day once, and it took me six months to close a half-million-dollar deal.

There is no set cycle. Today we have two points of customer focus. First, sovereign AI. We have a full solution that goes across the entire stack from the land to the models and everything in between.

Second, hyperscale infrastructure. We're working with a number of partners around the world who are looking for vast amounts of AI infrastructure, in the hundreds of megawatts at once. We're building the whole company around these two areas. **Jamie:** Forgive me, it's a really stupid question, but is this outbound outreach?

What does customer discovery actually look like? **Mahdi:** It's a relationship process. ** ** If you've been in the industry for long enough, you know the people. **Jamie:** I'd imagine the pool isn't enormous.

**Mahdi:** No. It's a small set of people. Through the industry, everyone knows everyone. **Jamie:** Are contracts on an annual basis?

Multi-year? **Mahdi:** These are contracts of five years plus. Some of them even longer. **Jamie:** So you're in it for the long haul.

**Mahdi:** Yeah. **Jamie:** Thank you so much, man. This was enormously enjoyable and thank you for being so transparent and open as well. It's super appreciated.

**Mahdi:** No problem. Thank you very much. Cheers.