Meta Doubles Down on AI Infrastructure in Louisiana While Pivoting to Paid Cloud with Muse Spark 1.1

Meta is executing a two-pronged strategy that pairs the world’s largest data-centre buildout with a fundamental shift in how it commercialises artificial intelligence. On Monday, the company announced that its “Hyperion” campus in Richland Parish, Louisiana, will now be scaled to 5 gigawatts of computing capacity – more than double the original 2 GW plan – pushing the total investment past $50 billion. Just days earlier, Meta unveiled Muse Spark 1.1, its first proprietary, paid-tier AI model, marking a clean break from its previous open-source approach and signalling an entry into the cloud and API business.

The Louisiana expansion, which broke ground in December 2024, has already generated more than $1.6 billion in contracts with local firms. Meta also plans to spend over $1 billion on roads, water and wastewater infrastructure in the region. The economic ripple effect is tangible: teachers in Richland Parish recently received annual bonuses of up to $50,000 – four times the prior year – funded by higher tax receipts from the construction. Meta is also contributing $5 million in scholarships to the Louisiana Delta Community College. Once fully operational, the facility is expected to support more than 1,000 permanent jobs.

Environmental groups have criticised the project’s enormous energy appetite. Earthjustice, an advocacy organisation, pushed for a review of the financing earlier this year but failed. Meta responds that it is covering all energy, water and infrastructure costs itself, and says its agreement with utility Entergy Louisiana will save customers more than $2 billion over two decades through avoided system upgrades.

On the product side, Muse Spark 1.1 – developed by Meta’s newly created Superintelligence Labs (MSL) – targets agent-based AI applications and complex coding tasks. It operates significantly faster than prior versions and is priced roughly 75% below comparable models from OpenAI and Anthropic. For Meta, the launch is a direct monetisation of the enormous capital expenditure that has underpinned its AI push. CEO Mark Zuckerberg has been candid that the company will now rent out spare computing capacity to third parties, turning infrastructure into a revenue stream that Wall Street has long hoped for.

Should investors sell immediately? Or is it worth buying Meta?

Alongside the AI pivot, Meta’s social-media franchise continues to expand. Threads has crossed 500 million monthly active users, fuelled by new features such as “Communities” and “Your Algo”, which hand users more control over their feed content. The platform is using AI-driven personalisation to deepen engagement, positioning itself as a serious rival in the short-messaging space.

Meta is also fighting a legal battle. The company is appealing a $375 million jury verdict in New Mexico that linked its social-media features – infinite scrolling, auto-play video – to addiction in young users. Meta disputes the findings and has pledged to defend its youth-safety practices, while acknowledging the case is part of a broader regulatory wave targeting platform design.

On the stock market, shares have been volatile but are trending higher. After closing at EUR 569.60 on Tuesday, down 1.33% on the day, the stock recovered to EUR 571.30 in subsequent sessions, trimming its loss to 1.04% from the prior close. Over the past 30 days, the equity has gained 11.63%. Still, at about 16% below the 52-week high of EUR 677.80 set in July 2025, the rally has not yet reclaimed peak levels. The 50-day moving average of EUR 519.76 and the 200-day average of EUR 547.45 both lie below the current price, confirming a medium-term upward trend. The relative strength index at 62.8 is approaching overbought territory but has not breached it.

Analysts see the cloud and API push as an essential hedge against the staggering investment trajectory. With Meta committed to spending $600 billion on US infrastructure over the next three years, creating a new profit centre from its AI compute capacity could help offset the margin pressure that such a buildout inevitably brings. The Louisiana data centre alone, the company says, is among the largest AI infrastructure projects on the planet, and investors will watch quarterly earnings for early signs that those watts are translating into revenue.