Meta is currently constructing a data center named Hyperion, which the company anticipates will provide its new AI lab with five gigawatts (GW) of computational power, as stated by CEO Mark Zuckerberg in a recent post on Threads. This announcement represents Meta’s latest strategy to outpace OpenAI and Google in the competitive landscape of AI development. Following its earlier recruitment of top talent to lead the Meta Superintelligence Lab, including former Scale AI CEO Alexandr Wang and former Safe Superintelligence CEO Daniel Gross, Meta appears to be focusing on the immense computational power required to train frontier AI models.
Zuckerberg mentioned that Hyperion’s footprint will be substantial enough to cover most of Manhattan.
Several media reports have cited a Meta spokesperson stating that Hyperion will be located in Louisiana, likely in Richland Parish, where Meta previously announced a $10 billion data center development. The spokesperson noted that Meta aims to bring two gigawatts of data center capacity online by 2030 with Hyperion, scaling up to five gigawatts in subsequent years.
Zuckerberg also emphasized that Meta plans to launch a 1 GW supercluster named Prometheus in 2026, positioning the company as one of the first tech firms to operate an AI data center of this scale. According to the spokesperson, Prometheus will be situated in New Albany, Ohio.
Read more | Meta AI app launches: Users can now manage daily AI interactions all in one place
Boosting competitiveness
Meta’s AI data center expansion is likely to enhance the company’s competitiveness against OpenAI, Google DeepMind, and Anthropic in its capacity to train and serve leading AI models. This initiative may also assist Meta in attracting additional talent, as individuals may be drawn to a company with the computational resources necessary to compete in the AI race.
Combined, Prometheus and Hyperion will consume enough energy to power millions of homes, potentially drawing significant amounts of electricity and water from nearby communities.
With federal officials’ backing, the AI industry appears set to absorb a large portion of America’s energy in the years ahead. Experts estimate that data centers could account for 20 percent of America’s energy consumption by 2030, up from just 2.5 percent in 2022. Without a rapid increase in energy production, this could lead to further challenges for local communities.
Major players race to enhance AI training capabilities
Tech giants across the AI landscape are launching ambitious data center projects to provide the unprecedented computational power required for training and deploying new generations of AI models.
OpenAI (and Microsoft)
OpenAI, in partnership with Microsoft, is planning a data center codenamed “Stargate,” which could cost up to $100 billion and require 5 GW of power, matching the scale of Meta’s Hyperion. Stargate is targeted for launch by 2028, with an interim 1 GW supercomputer being constructed in Wisconsin for debut as early as 2026. Additionally, OpenAI is expanding a site in Texas, projecting a total capacity nearing 1 GW, making it one of the largest AI-focused data centers worldwide. Katie Mayer from Microsoft/OpenAI emphasized, “These supercomputing systems are the lifeblood of our research. To carry out the work we’re doing at this scale and to create genuinely innovative AI capabilities, you need these systems.”
Google is also making significant strides, committing over $13 billion for new data center builds in the United States alone. This includes a $2 billion expansion in Indiana and $1 billion in Kansas City, with further investments planned in Northern Virginia and Iowa. Moreover, Google aims to run its AI and cloud data centers entirely on carbon-free energy by 2030 and is leveraging AIOps to automate and optimize workflows across its 19 global data center campuses.
Anthropic
Meanwhile, Anthropic is targeting $100 billion in investments to build next-generation data centers by 2027, intending to support the training of ever-larger “frontier models.” The company anticipates $10 billion data center projects as early as 2026 and primarily leverages Amazon and Google’s cloud AI superclusters for its operations.
Amazon
Amazon has invested over $4 billion into Anthropic, positioning itself as a preferred AI infrastructure provider rather than developing consumer-facing models like ChatGPT. Amazon Web Services (AWS) is significantly expanding its data center capabilities globally, hosting massive workloads for other AI startups.
Rising power demands
As these data centers proliferate, the rising power needs are a concern, as next-gen AI data centers are projected to consume energy equivalent to major cities. Data centers alone could account for 20 percent of U.S. electricity use by 2030, up from 2.5 percent in 2022. Local communities are already feeling the strain from projects like Meta’s in Georgia, which have led to resource depletion. In response, all major players are investing in infrastructure that combines advanced cooling, renewable energy, and grid partnerships to mitigate their ecological footprint.
