For our next chat we're joined by Arkady Volozh, I do hope that I said that correctly. Arkady Volozh, Founder and CEO of Nebius, with Elaine Moore, Tech Comment Editor at the Financial Times. Please give them a warm round of applause as they make their way on stage.
Thank you very much. Thank you. Hello everybody, lovely to be with you for this fireside chat. Arkady has had a fascinating month and a fascinating couple of years. Maybe let's start off—you have described Nebius as a startup or like a startup, but most startups don't have a thousand employees and two billion of capital and the sort of exciting history you've had. For those who might not know the history of Nebius, can you give us a quick run through?
So we start with history—a brief history. There was a company called Yandex, which I ran for 26 years. And then when the war started between Russia and Ukraine, half of the management and a thousand engineers left the country and actually started from scratch. And two years, two and a half years later, we totally separated corporately—we sold the Russian asset and started building a completely new company with several projects which were initially created to run on Western markets. And the main project was this creation of infrastructure for AI, actually—a network of clusters with GPUs, with processors on which all the models are trained and then run. And since we had some of the assets—one of the data centers was built outside—it came with us. And the team, actually, which we had in the previous life—they built the infrastructure of Yandex, hundreds of megawatts of data centers. They have this experience. We just created a new stack of technologies, built it on our existing data center, expanded. Now we have more data centers—we have a data center in Paris, we're building several new ones, we are launching several new ones soon in the US, and then we'll be building more and more. And now we are on this market—we provide tools to people who develop the models, train the models, run applications. We're a technology company for other technology companies. We're not B2B, we're T2T—tech to tech.
You built Yandex, which was called the Russian Google, and now you're building another company. It's been described as Yandex 2.0, but it's not, because you had this 5.4 billion sale—so you had the total division of Russian assets. Does it feel like you're building a startup again, or does it feel like something different?
It feels like a startup in the sense that we absolutely don't know what will be built in the end. We know where we're going, but how it will shape up in six months, in six years, we'll see. We're pretty sure about ourselves that we can deliver, we can do something, because we already did something before. But we don't have an existing business model yet—we need to build it. We didn't have a company, now we're building it. But yes, it's not just a startup—it's a startup with a lot of memory and a lot of knowledge in it. And yes, we kind of sold our Russian asset, which in its best time was a 30 billion dollar listed company—we sold it for five-something, two and a half billion in cash. And we have this cash, which makes us a pretty unusual startup in general, and a pretty unusual player on this specific market. And this specific market is the market of companies which provide specialized infrastructure for AI.
And so the Ukraine-Russia conflict led to the suspension of Yandex shares, and there was a period of time where there was no trading. Then you had the sale, and very recently there's been the relisting of shares. I wonder if you can tell us what that's like, and also, did you consider staying as a private company? Why relist?
Well, actually, more than two weeks ago we became—we stayed a NASDAQ-listed company, but NASDAQ had frozen all the trade because we had Russian assets and it was against their rules. As soon as we separated, we came back, said that we're ready, and then two weeks ago they said, 'Good, you're compliant, you go public next Monday at 9:00.' No warning—we had one working day to prepare the market for that. We created a nice press release, that's all we could do, and we started trading. And so far it goes well. The market forces work. Over these two weeks we had a very big volume of trade, obviously, because the old shareholders who invested in a completely different company and completely different market, probably didn't have mandate to stay in a B2B AI infrastructure provider—they exited, some new shareholders bought their stakes, and now we're still public—we have more, I don't know, 10,000 shareholders or more. The price was quite volatile in the beginning, now it's civilized a little bit, and we're a public company.
And is that—are you planning to raise, do you want to raise more capital? Because you're in this expansion mode, it's a capital-intensive—
It is important to be—actually, and we are lucky to be public because our sector, AI infrastructure, it stands on two pillars. One is technological stack, and another is capital. It's a very capital-intensive business—you need to buy, build data centers, you need to buy a lot of GPUs and then gradually amortize them. We're pretty sure about our technological skills, and among our peers—independent cloud providers, AI infrastructure providers, sovereign AI clouds—we are, as of today, the only public company. First of all, we are the only ones to sit on two billion—we have our own cash, which is free capital for us, and it's very useful. We can use it and build really big infrastructure using the first two billion we have. And then being public gives you cheaper access to many capital instruments—you can sell your stock, you can raise bonds or convertibles. We have better access to cheaper capital. So good technological stack allows us to be efficient and to have low prices or higher margins of our product, and cheaper access to capital gives us very important advantage.
And AI infrastructure is an area that has been dominated by some big US companies. I think you're the only European-based AI infrastructure full-stack company. Are you—are you thinking of that? Your clients are here, are you targeting Europe in particular?
We're not the only ones, but we're probably one of the largest in Europe, yes indeed. Well, still, we're enjoying being in Europe because we have some data centers here, because we were lucky to get access on early stage to some green fields which we're building. And Europe still had—this year, unlike in the US, US was taken, because Europe was slower, which was very lucky. We were lucky to be here. But our business—majority of our clients is in the US, because our primary clients, majority of our clients, are so-called AI native startups, and I don't know, 80, 90% of them are somewhere in the Valley. And these are our main clients. Of course we have European clients as well, companies like Mistral and others. But if you follow the market, the market today is there. It will change, definitely. AI will proliferate. Now we are talking about AI native startups—people who create models and create applications on AI. But then it will be more and more corporate market, and then of course Europe has huge corporate market. But as of today, our customer base is there in the US.
What's your view of the European tech startup sector? Is it—is there more that could be done? Is it being held back by too much regulation?
