Goal CEO Mark Zuckerberg announced Monday that the company will launch its first AI data center, called Prometheus, in 2026, according to a post on Threads.
Goal It is also building a second supercluster called Hyperion, which is expected to grow to five gigawatts within several years. Zuckerberg mentioned that Meta is working on several additional data center clusters, including one that will encompass a portion of Manhattan.
This data center expansion is part of Meta’s broader effort to develop superintelligence, a category of AI that the company differentiates from artificial general intelligence.
What Zuckerberg announced about the scale of AI investment from Prometheus, Hyperion and Meta
Zuckerberg’s post outlined a number of upcoming developments. Prometheus is expected to come online in 2026, while Hyperion will expand to five gigawatts over several years.
Meta is building multiple superclusters, one of which is planned to cover a large portion of Manhattan’s footprint. The company intends to finance these projects through its internal cash flow and commercial operations, avoiding external financing sources.
Zuckerberg mentioned: “We are also building many more Titan clusters. Just one of them covers a significant portion of Manhattan’s footprint.” The reference to Manhattan-scale infrastructure underscores the large physical and energy requirements needed for AI training and inference these days.
Meta has announced plans to invest hundreds of billions of dollars in computing infrastructure aimed at developing superintelligence. The company clarified that this financing will come from its commercial operations, taking advantage of the capital generated by the current strong performance of its core business. The scale of the investment reflects Meta’s estimate of the computing power needed to achieve superintelligence.
Alexandr Wang, Chief AI Officer at Meta, said: “For our superintelligence effort, I am focusing on building the most elite and talented team in the industry. We are also going to invest hundreds of billions of dollars in computing to build superintelligence.”
How Meta Frames frames superintelligence versus general AI and its strategy
Meta positions its work on superintelligence as opposed to general AI. According to Wang and the company, general AI refers to systems trained to perform a variety of tasks, while superintelligence describes systems that exceed human capabilities in broad domains. This distinction influences how Meta justifies its infrastructure investments.
The superintelligence framework also aligns with Meta’s competitive stance against companies like OpenAI, Anthropic and Google, each of which uses different terminology to describe the cutting edge of AI development.
The Prometheus and Hyperion announcements complement other Meta AI initiatives, including Muse Spark 1.1, a coding-focused model now in public preview in the US with competitive pricing; Muse Image, an image generation model that uses public Instagram profiles by default; Meta Superintelligence Labs, the internal team that develops these models; and Ambient Meta AI, which provides assistant functions in all Meta applications.
Meta’s infrastructure expansion supports its push into agent AI, coding tools, imaging, and consumer-facing AI features. Without extensive training and inference capabilities, Meta cannot compete at the forefront with companies like OpenAI, Anthropic, and Google.
Talent drive, energy and infrastructure concerns, and what users should watch next
Meta has been actively recruiting AI talent and offering compensation packages reportedly reaching hundreds of millions of dollars for senior engineers. This talent acquisition strategy supports the company’s infrastructure investments. Wang has stated that Meta aims to build the most elite and talent-rich team in the industry.
Meta’s approach differs from OpenAI, Anthropic and Google, which have historically relied on smaller, mission-focused teams rather than large-scale hires to build broad AI research and engineering capabilities.
Mention of the Manhattan-scale project and the five-gigawatt Hyperion cluster raises important questions about energy use and infrastructure, including where the energy will come from, how grid stability will be maintained, whether local communities will absorb rate increases to finance construction, and how environmental impacts will be assessed.
These concerns align with recent regulatory actions. New York has suspended permits for data centers of 50 MW and larger while it develops new rules. Maine attempted to implement a full state moratorium earlier this year, but was vetoed. President Trump has called for Big Tech to take responsibility for data center energy consumption.
Meta’s planned five-gigawatt Hyperion cluster is 100 times larger than New York’s 50 MW threshold for data center permits, indicating the scale of infrastructure the company plans to develop. How Meta obtains the necessary permits, energy agreements and community benefits will influence whether its timeline remains feasible.
For users following Meta’s AI development, Prometheus is expected to be operational in 2026, and its impact on Meta’s AI capabilities will become evident through enhanced product features. Hyperion will be rolled out gradually over several years and capacity will gradually increase.
There are additional superclusters in development, but their names and timelines have not been announced. As construction continues, community-level effects such as grid strain and environmental impacts are likely to emerge.
For users comparing Meta’s AI products to competitors: Meta’s investment in infrastructure indicates that its AI offerings will continue to expand on the frontiers of technology. Unlike many AI companies that rely on infrastructure partnerships, Meta builds its own data centers.
This approach can allow for faster iteration and broader access to models, as seen with releases like Muse Spark 1.1 under the Apache 2.0 license, which could benefit from Meta’s significant infrastructure investment.
Meta’s data center developments come amid a shift in the industry, where infrastructure has become a key competitive factor. Big players like Microsoft, Google, Amazon and Meta operate large-scale data center networks. Meanwhile, emerging competitors like Anthropic and OpenAI rely on partnerships with cloud providers like Amazon, Nvidia and Microsoft for access.
The race for infrastructure capacity increasingly determines which companies can operate at the highest scale. Meta’s Hyperion, designed to scale up to five gigawatts, positions the company among the largest hyperscalers for AI training and inference.
Prometheus is expected to be ready in 2026. Hyperion continues to be built for several years and additional superclusters are being developed without specific timelines. Interested parties can follow official Meta announcements for updates on the infrastructure rollout and its effects on AI products. Local news sources where Meta is building data centers are also likely to provide information on community impacts and permitting progress.
