Meta Unveils $10B Indiana Data Centre Campus
Meta Platforms announces a 1,000 MW campus in Lebanon, Indiana, supporting approximately 4,000 construction jobs.
Meta Platforms has announced plans for a $10 billion, 1,000 MW data centre campus in Lebanon, Indiana - its second major facility in the state and one of the largest single-site investments in Meta's history. The campus will support Meta's expanding AI training infrastructure, including the next generation of large language models (successors to LLaMA), recommendation systems that power Instagram and Facebook's feeds, and the computational requirements of Meta's growing AI assistant products.
At 1,000 MW (1 GW), the Lebanon campus would consume roughly the same amount of electricity as a city of 750,000 people. This staggering power requirement illustrates the scale disconnect between AI infrastructure and traditional commercial real estate - a single data centre campus rivaling the electricity consumption of a mid-size city. Meta's total infrastructure investment in Indiana alone now exceeds $15 billion across its two campuses, making it one of the largest corporate investors in the state's history.
The Lebanon campus will create approximately 4,000 construction jobs during the multi-year build-out and several hundred permanent operational positions. Indiana has actively courted data centre investment with competitive incentive packages including property tax abatements, sales tax exemptions on equipment purchases, and streamlined permitting. The state's central US location provides low-latency connectivity to both East and West Coast markets, while Indiana's power costs (approximately $0.07/kWh from a mix of coal, natural gas, and growing renewable sources) are competitive with major data centre markets.
Meta's infrastructure strategy is the most concentrated among the major hyperscalers. While Microsoft and Google spread their investments across dozens of markets, Meta has chosen to build fewer but significantly larger campuses. This mega-campus approach - building 500-1,500 MW facilities rather than 50-100 MW facilities - enables economies of scale in construction, power procurement, and operations that smaller deployments cannot achieve. The trade-off is geographic concentration risk: a single event (grid failure, natural disaster, regulatory change) could affect a disproportionate share of Meta's total capacity.
The $10 billion price tag reflects the rapidly escalating cost of data centre construction. At approximately $10 million per MW, the Lebanon campus is near the current global average of $11.3 million per MW (per JLL). However, the actual cost per usable MW of AI compute capacity may be significantly higher, as Meta's AI training facilities require specialised liquid cooling, high-density power distribution, and GPU-optimised network fabrics that add substantial cost beyond basic shell-and-core construction.
Related Articles
Liquid Cooling Deployments Triple as AI Server Densities Exceed Air Cooling Limits
Direct-to-chip and immersion cooling installations surged 200% year-over-year as rack densities for AI workloads push beyond 100 kW per rack.
12 min readTechnologyRecord $12B in New Subsea Cable Projects Announced for Trans-Atlantic and Trans-Pacific Routes
Google, Meta, and Microsoft are collectively investing over $12 billion in new submarine cable systems to support AI workload distribution across continents.
11 min readTechnologyAmazon, Google, Microsoft Announce Record AI Infrastructure Spending
The three largest cloud providers have collectively committed over $200 billion in data centre capital expenditure for 2026-2027.
10 min readNeed bespoke market analysis?
Our advisory team delivers in-depth research tailored to your investment and operational requirements.
Get Advisory Support