klamb_s/iStock/Getty Images Plus via Getty Images
Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- Anthropic will spend $50 billion to build its own data centers.
- They’ll be in Texas, New York, and elsewhere across the US.
- It could set new standards for AI labs trying to scale.
Anthropic will invest $50 billion in the construction of its own data centers, the company announced Tuesday — a move that could have ripple effects across an industry already racing to build AI infrastructure.
“Anthropic serves more than 300,000 business customers, and our number of large accounts — customers that each represent over $100,000 in run-rate revenue — has grown nearly sevenfold in the past year,” the company wrote in the announcement, citing the demand behind the expansion.
Also: AI will cause ‘jobs chaos’ within the next few years, says Gartner – what that means
The new facilities — slated to be built in Texas, New York, and other locations across the country that have yet to be disclosed — will be “custom built for Anthropic with a focus on maximizing efficiency for our workloads, enabling continued research and development at the frontier,” the blog post said.
An emerging trend
Tuesday’s announcement marks Anthropic’s first foray into building its own data centers. Vijay Gadepally, a senior scientist at the Massachusetts Institute of Technology’s Lincoln Laboratory and the cofounder of software company Bay Compute, said it could be a harbinger of a broader trend throughout the tech industry, as younger companies seek to secure sustainable sources of computing power.
Also: Does your chatbot have ‘brain rot’? 4 ways to tell
“This is, I think, definitely a direction that we’re going to see happening more,” Gadepally told ZDNET. “Three or four years ago, the biggest bottleneck was how many GPUs you could get your hands on, and that’s why a lot of these big model developers signed strategic agreements with big cloud providers or hyperscalers to get essentially guaranteed access.”
The new effort from Anthropic to build its own custom data centers, Gadepally added, “is the next logical progression of that: How much of the verticalization of compute can you get your hands on?”
Most AI startups don’t have the funds of an OpenAI or an Anthropic at their disposal, and will have to continue relying on partnerships and leasing agreements with third-party AI infrastructure companies. But for the small subset of developers who can afford it, following in Anthropic’s footsteps by building proprietary data centers could catch on.
Also: Google’s Private AI Compute promises good-as-local privacy in the Gemini cloud
“For companies that are training massive frontier models, there is a good chance that you’re going to see increasing verticalization,” he said.
Bubble fears and inflated promises
Anthropic’s new data center construction project will supposedly push the company toward achieving major new advancements with its technology.
“We’re getting closer to AI that can accelerate scientific discovery and help solve complex problems in ways that weren’t possible before,” company CEO and cofounder Dario Amodei said in the announcement. “Realizing that potential requires infrastructure that can support continued development at the frontier. These sites will help us build more capable AI systems that can drive those breakthroughs, while creating American jobs.”
Anthropic added that as the new sites come online throughout the next year, they’ll create 800 permanent jobs and 2,400 construction jobs, thereby helping to advance the goals laid out in the Trump administration’s AI Action Plan. Released this summer, the Action Plan focuses on ramping up infrastructure to maintain the US’s competitive edge in the AI race over other countries, especially China.
Also: As OpenAI hits 1 million business customers, could the AI ROI tide finally be turning?
Such proclamations have become a common refrain among AI developers at a time when fears about a potential AI bubble have been simmering. While billions in investor dollars are still flowing freely into AI, some experts worry that the technology won’t be able to financially deliver in the long run, and that the US economy could cave in as a result.
Last week, OpenAI wrote in a blog post that “superintelligence” — a hypothetical AI system that’s far more advanced than the human brain — could lead to “a world of widely distributed abundance” by, for example, assisting with scientific discovery and drug development.
Only time will tell if such promises can be turned into reality — and whether or not concerns about an AI bubble are justified.
Anthropic vs. other AI labs
As the stakes of the AI race have grown, so too has demand among tech developers for energy. It takes a lot of electrical power to fuel the supercomputers behind consumer chatbots like Claude, ChatGPT, and Gemini, and those in turn require huge quantities of water so that they don’t constantly overheat. Data centers also raise energy costs for residential areas surrounding them.
Also: Worried about AI’s soaring energy needs? Avoiding chatbots won’t help – but 3 things could
All of that energy-providing infrastructure is expensive, which means that deep-pocketed legacy tech companies like Meta, Amazon, and Google’s parent company Alphabet have had an immediate advantage over the vast majority of startups trying to sell new AI products. But OpenAI and Anthropic — both of which are relatively young companies that were quick to launch extremely popular chatbots — have been two of the most successful exceptions to that rule.
(Disclosure: Ziff Davis, ZDNET’s parent company, filed an April 2025 lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
Founded as a nonprofit in 2015, OpenAI started receiving billions of dollars in capital from Microsoft four years later, which enabled it to rapidly build new models and invest in the requisite infrastructure to power them. OpenAI doesn’t own these facilities; however, it mostly leases them from developers specializing in data center construction. Though the company is restructuring towards an ambitious future, it still relies on partnerships to access the computing infrastructure that fuels all of its AI products.
Also: The key to AI implementation might just be a healthy skepticism – here’s why
Earlier this month, for example, OpenAI announced it would pay Amazon Web Services $38 billion for access to that company’s cloud computing services as it ramps up its efforts to build artificial general intelligence (AGI) — and, more immediately, hit the goals that it set for itself when it kicked off Project Stargate, a collaborative data center construction project with the government and other big tech firms, in January.
Thus far, Anthropic has pursued a similar course, primarily tapping into data center architecture provided by its two biggest financial backers, Alphabet and Amazon. But the company has grown explosively since its founding in 2021 — with a reported valuation of $183 billion in September — thanks in large part to Claude’s popularity among enterprise users. Its options have therefore expanded.
