Hello everybody, welcome back to C&F Talks. Today I have the great pleasure of speaking with Karl Havard, who is the Chief Commercial Officer at Nscale. And Karl's been around in this business for some 25 years, previously held senior leadership roles at AWS and Google, and also has built leading startups, so he's well-versed in this area. And Karl's going to be speaking at our National AI Policy Infrastructure and Skills Summit, which is being held in London on the 12th of November. Karl, welcome.

Thank you, Maurice, good to be here.

Great to have you with us.

Now, we know that Nscale, and you personally, welcome the ambitions of the AI Action Plan from the very start. And obviously, there have been some great announcements recently. What do you think has been one of the most significant achievements, perhaps, since it was launched earlier this year?

Yeah, it was great to be part of that, because I think it was January the 13th, if I recall correctly, at the beginning of the year, where Keir Starmer, Peter Carl, also Matt Clifford, launched from UCL, I think it was, the Opportunities Action Plan. And we were pleased to get a mention within that, to be seen at the time as an upcoming AI company coming out of the UK. So, there was a lot of work to get ourselves noticed at that particular time, to talk about what our ambitions were coming out of the UK, offering sovereign AI to UK and international enterprises.

You then fast forward six months, and at London Tech Week, Keir Starmer's on stage, Jensen Huang, CEO of NVIDIA, joins him on a panel with Poppy Gustafsson, and starts to talk about what progress had been made in that six-month period. What's really interesting from that is, it's quite rare to see probably a UK leader or a governmental leader get behind an AI initiative.

But I think Jensen Huang did say, it's great, the UK has a fantastic heritage, both for the best universities in the world, the best AI brains in the world, as well as going back to the 50s and Turing times, and being at the forefront of this new industrial revolution, call it that, but we lacked something. And the thing that we lacked was the AI infrastructure in the UK to help organisations and start-ups and research institutions deploy. And there was still a reliance on the great companies like AWS, Google, Microsoft, etc.

So, you know, then fast forward a few more months, and we haven't solved that problem, but we are well on our way, as Nscale, with the investment we've achieved, to actually building out this AI infrastructure for the UK. And in sync with the UK government's AI growth zones as well. So, our aim is to deploy various clusters of NVIDIA technology into those areas.

So, progress has definitely been made. If I'm honest, that has mainly been from the commercial sector. I still think, despite all the support we're getting from UK government, and there's been obviously a change of ministers in the past month or so, we still have yet to see UK government investment into this. But we know that's just around the corner. But yeah. So hopefully that answers your question. Quite long, but yeah.

Yeah, it's a very good answer to the question. Interesting what you say. It's mainly on the commercial side. Some work yet to do perhaps on the government side.

And of course, you're building the UK's largest AI supercomputer with Microsoft at the campus in Loughton. So how is that going to...I'm interested in what is the utility of that for British companies, and how does it support the UK economy more generally?

Yeah, it's a good question. I mean, it's really interesting when we start to quote the largest this or the biggest this, et cetera, in a particular time. I mean, at this moment in time, we believe it will be.

We're probably looking at deployment, though, at the beginning of 2027. So, when we reach then, who knows if it's still the largest. But the great news is it's a UK data centre. We're building... So obviously we're creating UK jobs around that. We're bringing investment from overseas into the UK within it.

And Microsoft is going to house their supercomputer within the Loughton facility, which means they will be bringing customers themselves to it, as well as themselves investing. They'll be serving AI on a UK sovereign platform to their customers. And then we will be working with Microsoft as a service provider, which means the revenue that we generate from those customers will stay within the UK. And then we can then reinvest that into other areas accordingly.

So, we're super excited about being able to house someone as established and as big as Microsoft on a UK facility, which I think it's a vote of confidence in two things. It's a vote of confidence in the UK itself. And they see it as this Goldilocks position, which Jensen Hange said, but also a vote of confidence in someone like Nscale, a two-year-old organisation which seems to be doing the right things at the moment. So, yeah.

Absolutely. Yeah, I've forgotten for a moment that Nscale was essentially quite young.
You've achieved this sort of kind of pivotal role incredibly quickly, haven't you?

Yeah, it's sort of... It doesn't feel quick; I have to say that. You know, we've been immersed in this for a couple of years, but in the last month or so, it feels like coming out of nowhere. But I think what we've done at Nscale is we looked at the supply chain when it came to AI infrastructure.

And the world was focused on; can we get the GPUs in place in time? And that was looking at the supply chain of the manufacturers. Our view was that will sort itself out.

A machine will emerge and there'll be a regular release of AI infrastructure supported by great OEMs like Dell. And the issue will be places to put. And in order to house this AI infrastructure, as well as physical data centres, you also need access to power.

And therefore, what Nscale did, say back in the early days, was to make sure we had that foundation in place. So, arranging power agreements, not just in the UK, but up in Norway where we have Blomfield, one of our flagship sites as well, making sure we had access to good power at the right rates and also the planning permission to build and extend. So, we have gone through that process, and I think proved right that the supply chain of AI infrastructure has now got fairly slick.

The lead times were anything up to 12 months or more. I think lead times are now down to three months. And we're now in a position where we can secure contracts, build out the AI infrastructure and have a place to put them there. So, I think that has helped us get to a place faster than maybe others have been able to do.

Yeah.

I mean, that's quite an achievement because certainly the access to power, sufficient power for what you're doing, that must be a challenge. People talk more generally about the grid here and its capacity to deal with the various demands on it, whether that be renewables, whether that be electric cars. And now, of course, AI data centres, which are very large consumers of electricity. But was that the most difficult aspect of doing what you're doing?

I think the most challenging aspect of what we're doing, I don't think you can pick one element out. It's the combination and the synchronicity of things that happen together. So having planning permission in sync with network connectivity, in sync with energy, and then building the data centres or partnering with other data centres, winning the contracts for offtake. So, we have customers on board that will actually take that. I've got a fellow behind me, apologies for that. It's like a number of things.

So, where I think the challenge remains and will remain in the UK for some time is associated with getting renewable energy into the grid and trying to find a way to bring down the cost of energy in the UK. At the moment, it's around about 20 to 22 pence per kilowatt. And in Norway, it's about eight. So it’s a lot less.

So, when organisations, global organisations are looking to locate, that could be seen as a stumbling block for us. So I think data centre builds, us and other data centre providers are looking for ways to bring energy to the site, probably outside of the grid, and try to find a way to bring renewable energy, which is in abundance outside the UK, six gigawatts plus into the UK, into data centres by creating battery farms that can then store it directly, which then allows the data centre to maybe step off the grid for a few hours each day, which will then mean that that energy price on average can be reduced. What we would love to happen, though, is for data centres to be a catalyst to bring down energy prices and also bring renewable energy into the UK for the benefit of the citizens of the UK.

So, we're reliant then to work in sync with the national grid. But because of the way that the market is for AI and this, not an arms race, but it feels like there is a race often, and the ability to deliver quickly, it is forcing data centre builders to try and find alternative solutions. I would love us to be in sync with the national grid, so communities around the data centres would also benefit from that energy.

Yeah, and that'd be the perfect world, wouldn't it? Innovation, I mean, I suppose these problems drive innovation, don't they? But final question, because I think we're kind of running out of time.

I know that you together with OpenAI and NVIDIA are together establishing Stargate UK, which is going to be, as I understand it, an overarching infrastructure platform designed to deploy OpenAI's technology in the UK. Can you tell us a bit more about that?

Yeah, sure. I mean, it's sort of in sync with what we've done with Stargate Norway. So, Stargate Norway was our first sort of engagement with OpenAI.

So, building out 230 megawatts in Norway, deploying OpenAI, the first OpenAI installation into Europe, by the way, and for us to be able to be the engine that serves OpenAI across Europe. So, the same is very true for the UK. It's not at the same scale as is in Norway, but it's still 30,000 GPUs.

And the aim is to then on a UK sovereign AI engine to be able to have OpenAI on top of it, and we can then help inference the OpenAI solutions across the UK. And because it's predominantly inferencing, then that means latency plays a part. So for UK organisations that will use AI, OpenAI, and ChatGPT, and everything that goes business or personal, you know, going forward, once it's established and we've built it out in the UK, then we will be hopefully offering faster and quicker services to organisations in this kingdom that want to then use it.

So, it's a great vote of confidence again, you know, for working with someone like Sam Altman and OpenAI. And, you know, we're in a privileged position to be able to do that. But it goes back to that point, which, you know, being able to build out the UK's AI infrastructure, I think we've been able to establish ourselves, certainly with the Series B, you know, investment now as well, to be seen as that player that has the credibility to be able to do that. So, we're pleased to be seen in that way.

Yeah, I mean, it's an exciting time, it really is, Karl. We look forward to hearing much more about this from you and others at the conference.

For our viewers, if you'd like to join us to learn much more about this, the event is the National AI Policy Infrastructure and Skills Summit being held in London on the 12th of November. Have a look at our website www.cityandfinancial.com for further information.

Karl, I look forward to seeing you on the 12th of November.

Likewise, see you then.