I started investing in $NBIS in January 2025. Since then, dozens of smaller "neoclouds" have emerged. Almost every week, someone would ask me why I wasn't investing in one of these smaller players instead. The argument was almost always the same: Nebius was no longer as cheap as when I built my position, while these smaller companies supposedly offered significantly more upside. My answer was always consistent. I wouldn't bet against them. Demand for AI compute was exploding, capital was flowing into the sector, and investor sentiment was overwhelmingly positive. In that environment, even companies with questionable business models could perform exceptionally well. But I also consistently warned that this was an industry with unproven long-term unit economics and significant risks that, in my view, weren't always being adequately reflected in valuations. Buying GPUs is relatively easy when capital is abundant. Building a sustainable, differentiated AI cloud business is something entirely different. It's about having the right team, securing reliable access to power, developing infrastructure at scale, optimizing the total cost of ownership, maintaining access to affordable capital, building the right customer relationships, and ultimately generating attractive returns on invested capital over multiple hardware cycles. Not all "neoclouds" are created equal. Some are essentially intermediaries, relying heavily on third-party infrastructure, external financing, and a favorable pricing environment. Others are building vertically integrated platforms, with greater control over their infrastructure, hardware deployments, networking, software, and overall cost structure. Having a best-in-class engineering team, proprietary technology, established ecosystem relationships, and the ability to execute complex infrastructure projects at scale can create competitive advantages that are much harder to replicate than simply acquiring GPUs and plugging them into a power source. Of course, vertical integration comes with its own challenges, particularly the capital requirements and execution risks involved. But when executed well, I believe it can provide a meaningful edge through greater operational control, efficiency, and better long-term margins. The cost of capital is another critical differentiator. In such a capital-intensive industry, even a few percentage points in financing costs can dramatically change the economics of a project. And we still don't know what those economics will look like once the current supply-demand imbalance eventually normalizes. Being right about the future of AI compute demand doesn't automatically mean you'll be right about the long-term economics of every company trying to meet that demand. That's precisely why I chose to focus on the company I believed was best positioned to build a sustainable business over the long term: Nebius. Its engineering DNA, proven track record of execution, vertically integrated approach, capital allocation strategy, and focus on building a full-stack AI cloud platform were among the key differentiating factors behind my original thesis. I'm not going to go into detail about everything that differentiates Nebius from its competitors here, as I've been discussing those factors extensively since I first invested. The point of this post is simply to explain why I've always been so selective about where I invest within this sector. I never argued that every other "neocloud" would fail, or that their stocks couldn't outperform $NBIS in the short term. In fact, I consistently said the opposite. My point was that not all these companies would emerge as long-term winners, and I preferred owning the business I had the highest conviction in rather than chasing whichever smaller player appeared cheapest or had the most upside on paper. Now, with growing concerns about the sustainability of the neocloud business model and the availability of financing, the market is beginning to pay more attention to some of those risks. And frankly, I think that's healthy. It was never particularly healthy to see companies pivoting from completely unrelated industries into AI infrastructure, without experienced teams, and still managing to raise capital simply to acquire a few GPUs. If the sector begins to consolidate, I actually believe that could strengthen the position of the best-capitalized and most operationally capable players. Weaker competitors scaling back could mean less new capacity coming online, a longer-lasting supply-demand imbalance, and potentially better pricing and margins for those able to continue expanding. Over the long term, I believe the companies with genuine competitive advantages will be in a much stronger position. And to be clear, Nebius isn't immune to any of these risks. This is something I've emphasized in every single article I've written about the company: $NBIS, and this sector as a whole, isn't suitable for every investor. We're talking about an industry with significant uncertainties, where execution will be critical and many questions about the long-term economics remain unanswered. But nothing about the recent debate changes my original investment thesis. I never invested in Nebius because I believed every "neocloud" would succeed. I invested because I believed Nebius had the right differentiating factors and long-term vision to become one of the few sustainable winners in this industry. And I still do.

This is honestly insane. Marc Boroditsky, CRO of $NBIS, just revealed that the company has 1 GW of Vera Rubin GPU capacity coming next year, with significant customer demand already lined up. That alone could translate into tens of billions of dollars in ARR. 🤯

🚨 AIB Data Centers signs a contract with $NBIS for 50 MW of AI data center capacity in the U.S. The contracted capacity is supported by AIB’s previously announced 15-year Electric Service Agreement covering 65 MW of utility load at the site, with no significant additional electrical infrastructure upgrades required. AIB expects to deliver the capacity across two data halls. “Time-to-power is the binding constraint on AI infrastructure today, and AIB’s existing power position gave us a clear path to bringing this capacity online on a timeline that works for our customers. This agreement adds dedicated capacity in the southeastern U.S. for training and inference workloads.” — Andrey Korolenko, Head of Infrastructure at $NBIS