At a world forum obsessed with artificial intelligence and energy, one image cut through the jargon: a data center campus so vast it could match Manhattan in footprint, “miles long, miles wide, and very high.” The description, relayed by President Donald Trump after a private briefing from Mark Zuckerberg, crystallized just how outsized Meta’s next wave of infrastructure has become. What might once have sounded like a sci‑fi exaggeration is now a live proposal with a price tag and a political backlash to match.
Behind the spectacle is a hard strategic bet. Meta is racing to dominate AI by building some of the largest silicon clusters on earth, and Zuckerberg is signaling that only a Manhattan‑scale complex can deliver the compute he wants. I see that ambition colliding with a new era of scrutiny over land, water, and electricity, as communities and regulators begin to ask whether an AI arms race justifies a facility that could swallow an entire city grid.
The Manhattan‑scale vision that stunned Davos
When Trump recounted his conversation with Mark Zuckerberg at Davos, he lingered on the sheer physicality of the plan. He said Zuckerberg showed him a rendering of a data center the size of Manhattan, describing a structure that was “miles long, miles wide, and very high,” and attaching a projected cost of $50 billion. Coming from a real estate developer who has spent a career talking in superlatives, the awe in Trump’s retelling underscored how far beyond conventional industrial projects this proposal sits. It is not just another server farm, it is an attempt to build one of the largest silicon clusters ever assembled.
The concept has quickly become a talking point across climate and tech circles. Coverage of Mark Zuckerberg’s plan stresses that the facility would rival Manhattan in scale, with advocates touting its role in powering advanced AI and critics warning about the land and resource footprint. At Davos, Trump framed AI as “massive” and cast the project as a symbol of American technological might, recounting how Donald Trump and Zuckerberg discussed the energy build‑out required to keep such a complex running. The Manhattan metaphor has stuck because it translates abstract compute capacity into something anyone can picture: a digital city, humming day and night, dedicated solely to machine intelligence.
Meta’s AI arms race and the supercluster strategy
For Zuckerberg, the Manhattan‑size campus is not a vanity project, it is the logical endpoint of a strategy that treats AI compute as the defining resource of this decade. Meta has already signaled that it will spend “hundreds of billions of dollars” on huge AI data centers in the United States, with Meta founder Mark Zuckerberg explicitly tying that investment to out‑innovating rivals. Reporting on Meta’s infrastructure roadmap describes a lattice of AI “superclusters” designed to support the company’s most compute‑hungry research, including work at its Superintelligence Lab that aims to compete directly with OpenAI, Google, and Anthropic. In that context, a single campus the size of Manhattan becomes a keystone asset rather than an outlier.
Meta CEO Mark Zuckerberg has already floated plans for a massive AI data center that would match Manhattan in area, positioning it as a way to consolidate tens of thousands of GPUs and associated equipment into one hyper‑dense complex. One analysis of the project notes that Meta CEO Mark sees this as essential to keeping pace with competitors like OpenAI and Google, while another describes how, as the company builds facilities on a scale that rivals Manhattan itself, the challenge is not just generating intelligence but managing the physical sprawl of GPUs and cooling systems. As the company builds facilities that, As the analysis puts it, rival Manhattan in scale, Meta is effectively betting that whoever controls the largest, most efficient AI superclusters will control the next era of consumer and enterprise technology.
Local flashpoints: Louisiana, Arizona and the politics of scale
That global ambition is colliding with local politics in places that look nothing like Manhattan. In Louisiana, Meta’s plans to build a sprawling data center complex have already triggered Senate scrutiny, with lawmakers zeroing in on how little the company has disclosed about its long‑term fuel mix and emissions. Critics argue that Meta’s assurances about clean energy are vague and offer little reassurance in a state where petrochemical interests and climate risks already sit in uneasy tension. Environmental lawyers have gone further, with Earthjustice filing a motion that highlights significant changes to the proposed size of the Louisiana data center and argues that the public review process and environmental safeguards, which were already severely lacking, have not kept pace with the project’s expansion.
In the Southwest, water‑stressed communities are mounting their own resistance. In Arizona, a proposed data center project near Tucson has become the subject of a lawsuit that alleges the county hid a “bodyguard of lies” to mislead the public about the facility’s true intensity and its impact on water scarcity and rising energy costs. Residents and local officials are pointing to the way Jan’s legal filings describe a pattern of underplaying resource demands, a pattern they fear could be repeated if a Manhattan‑scale complex lands in a similarly fragile region. In Louisiana, Earthjustice’s motion, filed in Jan, underscores how quickly these projects can grow in scope, leaving communities scrambling to understand what they have actually agreed to host.
Energy, emissions and the new data center backlash
Behind each local fight is a broader reckoning with the energy appetite of AI. As electricity bills climb amidst a worsening climate crisis, advocacy groups argue it is unconscionable that Big Tech is plowing ahead with energy‑hungry data centers without binding guarantees on efficiency or clean power. A new campaign launched in Jan calls explicitly to “stop data centers now,” casting the Manhattan‑scale proposal as the logical extreme of a model that externalizes costs onto ratepayers and local ecosystems. Organizers single out Big Tech’s push for ever larger AI campuses as a driver of new gas plants and transmission lines that will lock in emissions for decades.
Investors and utilities are also starting to voice concern. Reporting on the data center boom notes that Google, Microsoft, Oracle and Apple are all racing to expand capacity, and that these tech titans, along with Apple, plan even heavier spending in 2026. In that context, Meta CEO Mark Zuckerberg has defended Big Tech’s cash burn on infrastructure, arguing that the industry is working closely with grid operators and regulators. Yet the same analysis warns that rising electricity rates and grid constraints are becoming the next major roadblock for hyperscale projects, with regulators demanding more transparency and monitoring of their true impact. When Trump talks at Davos about expanding energy production to feed AI, he is effectively acknowledging that a Manhattan‑size complex would be inseparable from a broader reshaping of national energy policy.
From superclusters to public consent
Inside Meta, the Manhattan‑scale campus fits into a narrative of relentless scaling. Reporting on the company’s AI infrastructure push describes how, Alongside Prometheus and Hyperion, Meta is also preparing a series of “titan” clusters to further increase total capacity, a naming scheme that leaves little doubt about the intended magnitude. One detailed account of Meta’s AI build‑out notes that Alongside Prometheus and is planning a 100 billion dollar supercluster program, with the Manhattan‑size facility likely serving as a flagship node in that network. Another analysis, by Colin Kirkland, frames these controversial data centers as Meta’s bid to gain a competitive advantage over OpenAI, Google, and Anthropic, particularly through work at the company’s Superintelligence Lab. In that framing, the Manhattan complex is less a one‑off marvel than the physical embodiment of a corporate strategy that equates size with intelligence.
More from Morning Overview