Good morning. It's really a pleasure to be here today with Arkady. As mentioned, he's the founder and CEO of Nebius, which is one of the largest independent AI infrastructure providers. Its headquarters and main R&D presence is in Amsterdam, with additional R&D hubs across the US, Europe, and Israel. Before launching Nebius, Arkady co-founded Russia's Yandex and ran it as CEO for 25 years. I first met you in Moscow, probably—I don't know—about 15 years ago, and since then we have intersected at tech conferences all over the world. I'm glad that you are here today at DLD at this crucial time in the AI market to explain what you're doing now. But I did want to take a moment for those of you who may not be familiar with Yandex to explain how Arkady got to where he is today. Yandex was founded in 1997, a year before Google, and it counted Esther Dyson, a longtime DLD attendee who I think is here today in the audience, as one of its early investors. And Yandex was really one of the rare Russian tech success stories. It became known as the Google of Russia, given that it sold products broadly similar to its US counterpart, including search, e-commerce, advertising, maps, transportation, and a lot more.
Now, while the Yandex primary market was Russia, the company went public on the NASDAQ in 2011 via a holding company based out of Amsterdam, or in the Netherlands, and that was followed by a secondary listing three years later on the Moscow Exchange. Yandex hit a peak market cap of $31 billion on NASDAQ in November of 2021. However, in the months that followed after the Russian invasion of Ukraine, NASDAQ put a temporary hold on the trading of Yandex, and so Yandex NV in the Netherlands divested its Russian assets—a process that took two long years and was very stressful—and the company renamed as Nebius. It's now trading on NASDAQ under a new ticker, and following this listing about a month and a half ago, it raised $700 million from Nvidia and venture capital firm Accel. Now, with more than one trillion dollars projected to be spent on AI infrastructure in the near term, and the fact that Nebius has been named as a launch partner for Nvidia's next-generation Blackwell platform, The Motley Fool said in a January blog post that it believes Nebius is positioned to have a, quote, monster year in 2025. So before we get to 2025, I do want to take a moment against that backdrop to ask you about your incredible wild ride in 2024. I mean, you divested the Yandex Holdings from Russia, you launched Nebius in July, you built new data centers in November, you relisted on NASDAQ, and acquired new investors by December. How does it feel to start over?
Well, it feels like a startup. We're a young startup—six months and seven days old. What generally happened: when the invasion started, half of Yandex top management and a thousand engineers left the country, and two years, two and a half years later, we started up as a new company with several businesses, the main, the largest of which is this AI infrastructure. Because we have with us all the engineers who used to build Yandex infrastructure—megawatts, hundreds of megawatts of data centers. We were one of the largest Nvidia clients, I think the largest outside of China. We kept these good relationships, and we realized that when we left, there was no ChatGPT, and then when this new wave of gen AI started, we realized that probably this is exactly the place where we're very much relevant, because we know what it is about, we know where it will be in a few years, and this is what we can do one of the best. And we decided that we will build this business, and this business is built on three big pillars. One is, of course, technology. It's AI infrastructure, and infrastructure means that it's technology and a lot of capital. So on the technology front, we have one of the best engineers in this field who worked together for decades, who know what they're doing, what they're building. We have no doubt that we know how to build this infrastructure, and we have the product now, and we have the data centers, and we're building very fast the new ones. That's the first pillar.
The second pillar: we have enough capital, and in this new market, there are not so many independent companies who have money to build this. We have a couple of billion dollars with us. It's not the 31 billion which it was in the past, but still a couple of billion dollars to start with. We raised some more from Nvidia and several dozen investment funds, including Accel. So we have the second leg, the second pillar, which is the capital, which allows us to build a lot of these new data centers. And we already built some last year and continue building very fast this year. It will be hundreds of megawatts of data centers, and in 2026, we're ready to build a gigawatt and more, providing there will be a demand. We know how to build the supply—technology and capital, we have it. And the third pillar, the third leg, is actually to be sure that demand is there—the go-to-market strategy. And that's why this is exactly what we're building now. Again, we're very young; we started to build it a couple of months ago, and we hope that this is actually our main focus for this current period.
So you have the AI talent—you brought out, as you said, about a thousand people, top talent from Yandex in Russia, you brought them outside. You have the data centers, you have the GPUs, but you also have another differentiator from other alternative infrastructure providers, and that is the software stack and being able to provide. So can you talk about that?
Well, the infrastructure is a thin layer which supports this whole new wave of AI—gen AI. It's a big wave, it's as big as—I don't know—the internet was 20 years ago. We're actually rebuilding the internet, that's how it feels these days. I still remember what it was in the '90s, 2000s, and this is the new wave with the same consequences. I think there will be a lot of business to build in this area.
What I was asking you about your differentiator and about the fact that you can go beyond, and so people who—
Yeah, this business is very much run by the big tech companies—the three, four, five, you name it. We used to be one of those companies in our past lives; we remember what it is. But now what we can offer is an alternative to these big tech companies. They are full-stack companies who actually build infrastructure, models, applications, everything there, and they are the major consumers of these technologies. But there is also the second part of the market—a lot of independent AI developers, and we're targeting those. And for those companies, independent like ourselves, startups, we know very well what it is to be outside of this ecosystem, to be a startup. We built a thin layer of infrastructure which they could use as an alternative—they actually compete with these big tech companies. And there is a whole bunch of these new companies, alternative to the big tech—they're called neoclouds now. And among those neoclouds, there is just a bunch of several companies who are really big, who can build at scale—just, I don't know, half a dozen or less. And we definitely will be one of those. And we're different a little bit from the rest of this class of neoclouds. One, because we have our own technologies—we're not buying from different places, we are not the integrator. We build the data centers ourselves, we build them very efficiently with less energy consumption. We build our own racks inside the centers—the computers and disks inside—which gives us even more efficiency. And we have our own software layer on top, a cloud. So we are a full-stack company which provides everything as a package, which makes us more efficient. And the second thing, as I said, the second pillar is capital. We have access to—we first of all have our own cash, and we're a public company, we have better access, cheaper access to capital. So combined, we hope we will be one of the most efficient alternative AI infrastructure providers, one of the most efficient neoclouds.
