Meta Platforms (META -0.53%) might be the most misunderstood artificial intelligence (AI) play in the market. Up 30% year to date through Sept. 8, the stock has quietly outperformed most tech giants while trading at a significant discount.
At 25.4 times forward earnings, Meta stock costs 33% less than Nvidia (NVDA -2.01%) despite building what could become the world's most powerful AI infrastructure. Mark Zuckerberg is orchestrating one of history's most aggressive pushes toward artificial general intelligence (AGI) -- and the stock market hasn't caught on yet.
Here's what investors need to know right now.
Image source: Getty Images.
The $72 billion bet Wall Street is missing
Meta plans to spend $66 billion to $72 billion in 2025, largely on AI infrastructure. That is the bet the market is still discounting.
What the money buys is city-scale compute. Hyperion in Louisiana is planned to scale to 5 gigawatts with a footprint "nearly the size of Manhattan." Prometheus in Ohio targets about 1 gigawatt coming online in 2026. Five gigawatts is enough to power roughly 4 million homes.
The financials can support it. In Q2 2025, revenue rose 22% to $47.5 billion, and free cash flow came in at a healthy $8.5 billion.
Meta also stood up Meta Superintelligence Labs in mid-2025 with a clear aim: a personal superintelligence that helps users achieve their goals. This is not a chatbot project. It is an AGI effort meant to touch every product, from Instagram to WhatsApp.
The talent war Meta is winning (and losing)
Meta invested about $14.3 billion for roughly a 49% stake in Scale AI, and Alexandr Wang, Scale AI's co-founder, joined to lead Meta Superintelligence Labs. Nat Friedman, former Microsoft GitHub CEO, co-leads and heads Meta's applied AI research.
NASDAQ: META
Key Data Points
Before an August hiring pause, Meta hired around 50 researchers from Alphabet, OpenAI, and Apple, with some offers reported in the nine figures over four years. Recent reported hires include Trapit Bansal, an early OpenAI researcher in reasoning, and Jian Zhang, a longtime Apple robotics lead, who joined Meta's robotics efforts in early September.
But there is turbulence. At least three researchers resigned within weeks, and two returned to OpenAI. Meta reorganized the lab into four groups within two months of launch. The churn suggests that money alone does not guarantee retention.
Even so, the departures may not matter. Meta's open-source Llama models are competitive on many benchmarks, and, most importantly, the company reaches 3.48 billion Family Daily Active People (DAP), giving it unmatched distribution when new AI products mature.
The valuation gap that shouldn't exist
Nvidia trades at 38 times forward earnings because it sells picks and shovels for AI. Meta trades at 25 times earnings while building the entire mine. The market assigns Nvidia's AI exposure a 50% premium, even though Meta's annual capital expenditures (capex) plan of $66 billion to $72 billion equals roughly two-quarters of Nvidia's quarterly data-center revenue.
The bear case focuses on Meta's failed metaverse bet. But AI is different. Every tech giant is racing toward AGI because the winner could dominate computing's next era. Meta has the capital, computing power, and crucially, the user base to compete with anyone.
When infrastructure becomes destiny
Meta generates enough cash each quarter to self-fund its AI build-out. OpenAI relies on outside capital. Anthropic leans on cloud partnerships. Meta says its teams have industry-leading compute per researcher, enabled by its scale and cash flow.
Bottom line: Meta offers rare AI upside at a reasonable valuation. The stock does not need AI to work; the core business already supports today's price. If the superintelligence push succeeds, today's valuation will look cheap.
The setup is asymmetric: limited downside from a highly profitable core, and meaningful upside if the superintelligence effort lands. The market is not pricing Meta like a front-runner for AGI. That gap is the opportunity, and the core reason this tech giant stands out as one of the best AI stocks to buy now.
