Anthropic, the artificial intelligence company behind the Claude chatbot, is reportedly in talks to significantly expand its computing capacity at a West Virginia data center campus. According to Bloomberg, the company is negotiating a second deal worth roughly $45 billion with Nscale, a developer of AI data centers, to rent computing power at the Monarch facility.
This would follow an earlier agreement announced in August, in which Anthropic already committed to a $45 billion arrangement to lease capacity at the same campus. The new talks reportedly center on a second building that could deliver 460 megawatts (MW) of computing power, using Nvidia's next-generation Vera Rubin chips.
What is Nscale and the Monarch campus?
Nscale is a relatively new player in the booming business of building and operating data centers specifically designed for AI workloads. The Monarch campus, located in West Virginia, is one of its flagship projects. These facilities are essentially massive warehouses filled with thousands of specialized computer servers that train and run large AI models like Anthropic's Claude.
The scale of these deals is staggering. A $45 billion contract is among the largest ever signed in the data center industry, rivaling some of the biggest infrastructure projects in the world. For context, that's more than the annual GDP of many small countries.
Why is Anthropic spending so much on computing power?
Anthropic, like other leading AI labs, is in a race to secure enough computing power to train and deploy its models. AI models require enormous amounts of processing capability, and the demand for these chips has far outstripped supply. By signing long-term contracts with data center developers like Nscale, Anthropic is essentially locking in access to scarce servers and electricity for years to come.
This is part of a broader trend in the AI industry, where companies are spending billions to build out the physical infrastructure needed to support the AI boom. The AI data center boom has been a major driver of growth for companies like Schneider Electric, which makes cooling and power equipment for these facilities.
The role of Nvidia's Vera Rubin chips
The new Monarch building would reportedly be powered by Nvidia's next-generation Vera Rubin chips, which are expected to be even more powerful than the current H100 and H200 processors. These chips are designed specifically for AI workloads and are in extremely high demand. Nvidia's dominance in this market has made it one of the most valuable companies in the world.
However, the availability of these chips is not guaranteed. Nvidia has faced supply chain constraints, and the timeline for Vera Rubin's release is still uncertain. If the chips are delayed, it could push back the completion of the new Monarch building.
Financing and power: the big hurdles
These massive projects are not without risks. First, they require enormous amounts of financing. Nscale will need to raise billions of dollars to build the new facility, and that money may come from a mix of debt and equity. In a similar vein, Anthropic has already offered Blackstone a 10%+ return to fund a $9 billion chip lease debt, showing how these companies are using creative financing to secure the hardware they need.
Second, and perhaps more critically, these data centers need a massive amount of electricity. A 460 MW facility is enough to power around 350,000 homes. Getting that much power hooked up on time is a major logistical challenge, and delays in grid connections have plagued many data center projects across the country.
What it means for investors
For everyday investors, this news is a reminder of the enormous scale of the AI infrastructure buildout. Companies like Anthropic are spending tens of billions of dollars to secure computing power, and that spending is flowing through to a wide range of companies: chip makers like Nvidia, data center developers like Nscale, and the utilities and equipment providers that support them.
The demand for copper and other raw materials is also being boosted by the construction of these facilities. And the broader trend is lifting the fortunes of companies like L&T Technology Services, which provides engineering services for data center projects.
However, investors should also be aware of the risks. These deals are complex and depend on financing, construction timelines, and the timely delivery of chips. If any of these falter, the projects could be delayed or scaled back, which would impact the companies involved.
For now, the talks between Anthropic and Nscale are still in the negotiation stage, and there is no guarantee a deal will be reached. But if it goes through, it would be a clear sign that the AI compute boom is far from over.
