Meta is essentially building a small city's worth of electrical infrastructure
In Lebanon, Indiana, Meta has broken ground on a data center campus representing more than $10 billion in investment — a physical monument to the invisible computational arms race reshaping modern technology. Spanning 4 million square feet and designed to draw one gigawatt of power, the facility is less a building than a declaration: that the future of artificial intelligence will be won or lost in the unglamorous work of steel, electricity, and cooling systems. The Midwest, long peripheral to the geography of technological ambition, now finds itself at the center of a new kind of infrastructure frontier.
- The AI compute race has moved from boardrooms to bulldozers — Meta's $10 billion groundbreaking in Indiana signals that the infrastructure competition is now physically, irreversibly underway.
- One gigawatt of dedicated power — enough for 750,000 homes — will be consumed entirely by machines running the mathematical operations behind Meta's AI models and its billions of daily social media users.
- With rivals like Google and OpenAI aggressively expanding their own compute capacity, every month between groundbreaking and operational status represents a potential gap in Meta's ability to train larger models and serve users at scale.
- Meta's choice of Lebanon, Indiana over coastal tech hubs reflects a calculated pivot toward regions where land is available, power grids are accessible, and the infrastructure buildout can move faster.
- The facility is expected online by late 2027 or early 2028 — a timeline that will test whether Indiana's power grid, labor market, and real estate can absorb this scale of investment without creating the very bottlenecks Meta is racing to avoid.
Meta broke ground this week on a data center campus in Lebanon, Indiana, committing more than $10 billion to a facility that will span 4 million square feet and deliver one gigawatt of power. The investment ranks among the company's largest AI infrastructure bets, designed to support both its expanding artificial intelligence work and the computational backbone keeping Facebook, Instagram, and WhatsApp running for billions of users.
The scale reflects a deeper reality: as AI models grow more sophisticated, the companies building them face a hard physical constraint — they need enormous, reliable electrical power and the land to house it. Meta's decision to anchor this investment in the Midwest, rather than in a coastal tech hub, signals a strategic calculation about where power is available and where construction can move quickly. Indiana has been actively courting this kind of investment, and Lebanon offers the combination of grid access and open land that facilities of this size demand.
Rachel Peterson, Meta's vice president of data centers, indicated the campus should reach operational status by the end of 2027 or early 2028. That timeline carries competitive weight. In a landscape where Google, OpenAI, and others are racing to secure compute capacity, months of advantage in training larger models can matter enormously.
The one-gigawatt figure is worth sitting with — it is the electrical equivalent of powering a small city, redirected entirely toward machines performing mathematical operations on data. What Meta is building is not just a facility but a piece of national computational infrastructure, and the broader question it raises is whether the regions absorbing this wave of investment can scale their power grids, labor forces, and supply chains fast enough to keep pace with the ambition behind it.
Meta broke ground this week on a data center campus in Lebanon, Indiana that will cost the company more than $10 billion to build—a sum that places it among the largest artificial intelligence infrastructure bets the social media giant has made. The facility, which will sprawl across 4 million square feet, is designed to deliver one gigawatt of power, enough to fuel both the company's expanding AI ambitions and the computational backbone of its core social media platforms.
The scale of the commitment reflects the intensity of the infrastructure race now underway among the largest technology companies. As artificial intelligence models grow more sophisticated and computationally hungry, the companies that build them face a fundamental constraint: they need enormous amounts of electrical power and cooling capacity, delivered reliably and at scale. Meta's decision to anchor this investment in Indiana, rather than in a coastal tech hub or a region already dense with data center infrastructure, suggests the company is thinking strategically about where power is available and where land can accommodate the sprawling physical footprint these facilities require.
According to Rachel Peterson, Meta's vice president of data centers, the Indiana campus is expected to become operational by the end of 2027 or in the first months of 2028. That timeline matters. In a competitive landscape where companies like OpenAI, Google, and others are racing to secure compute capacity, being first to operational status can mean months of advantage in training larger models and serving users at scale. The groundbreaking itself—happening on a Wednesday in February—signals that Meta is moving past the planning phase and into active construction.
The one-gigawatt capacity is a telling detail. A gigawatt is enough power to supply roughly 750,000 homes. Meta is essentially building a small city's worth of electrical infrastructure, dedicated to machines that will run mathematical operations on data. That power will support both the company's AI research and development efforts—the work that goes into building and improving the large language models and image generators that have become central to Meta's competitive positioning—and the everyday infrastructure that keeps Facebook, Instagram, and WhatsApp running for billions of users worldwide.
The Indiana location also carries practical implications. The state has been actively courting data center investment, and Lebanon sits in a region with access to reliable power grids and available land. For Meta, the decision to invest here rather than expand existing facilities elsewhere suggests the company has calculated that this location offers the best combination of cost, power availability, and speed to deployment. It is a bet on the Midwest as a center of computational infrastructure, a shift from the coastal concentration of tech investment that has defined the industry for decades.
What the investment signals, more broadly, is that the artificial intelligence infrastructure race is no longer theoretical. It is concrete, it is expensive, and it is happening now. Meta is committing $10 billion and years of construction time to ensure it has the computational horsepower to compete. Other companies are making similar bets. The question that follows is whether the power grid, the real estate market, and the labor force in regions like Indiana can absorb this wave of investment without creating bottlenecks that slow everyone down.
Citas Notables
Rachel Peterson, Meta's vice president of data centers, said the Indiana campus is expected to become operational by the end of 2027 or in the first months of 2028.— Rachel Peterson, Meta VP of Data Centers
La Conversación del Hearth Otra perspectiva de la historia
Why Indiana? Why not expand in California or Virginia where data centers already cluster?
Power and land. Indiana has cheaper electricity and open space. When you need a gigawatt and 4 million square feet, you go where the grid can handle it and the land is available.
A gigawatt sounds enormous. What does that actually mean for what Meta can do?
It's enough to run the training and inference for massive AI models, plus keep Facebook and Instagram running for billions of users. It's not one or the other—it's both, simultaneously, at scale.
The timeline says operational by end of 2027. Why does that matter so much?
Because every month of delay is a month your competitors are training models you're not. In AI, compute capacity is a bottleneck. Being first to operational means you can iterate faster.
Does this mean Meta thinks it's losing the AI race?
Not losing—but it's not winning yet either. This is Meta saying: we're serious about competing. We're willing to spend $10 billion and wait two years to prove it.
What happens to Indiana?
Construction jobs now, permanent jobs later. But also: power demand spikes, real estate prices shift, and other companies will watch to see if this works before they decide whether to follow.