Meta’s $145B AI Spending Surge: Why Investors Dumped Stock After Watermelon’s GPT-5.5 Claim

Posted on by

(SeaPRwire) –   By: Reginald Vance

Meta’s stock fell 4.9% Friday, even as its superintelligence chief claimed the upcoming Watermelon AI model matched OpenAI’s GPT-5.5. Investors didn’t buy the progress update. Instead, they fixated on the company’s revised 2026 infrastructure spending forecast. Meta now plans to spend $125–$145 billion, up from an earlier $115–$135 billion. The jump stems from rising component costs and expanded data center plans. Watermelon uses an order of magnitude more compute than April’s Muse Spark (codenamed Avocado), which means far higher operational expenses. Wang’s claim wasn’t independently verified, and Meta declined to comment. OpenAI didn’t respond either. Investors have good reason to be skeptical—Meta has spent years playing catch-up to OpenAI, Google, and Anthropic without public market leadership.

Meta’s spending isn’t just about talent—though it’s offered top AI researchers hundreds of millions to join. The bulk goes to chips and data centers, the backbone of AI compute. An order of magnitude more compute for Watermelon translates to tens of thousands of additional GPUs, a resource already in global short supply. NVIDIA, the dominant GPU maker, has seen demand outstrip production for months. TSMC, which manufactures critical AI chip components, is also operating at near capacity. Meta’s increased spending will only intensify competition for these scarce resources, driving up prices further. Wang also teased major coding and agentic gains in an upcoming Muse Spark update, but even those incremental improvements don’t offset the long-term cost burden. OpenAI released GPT-5.5 in April and debuted GPT-5.6 late last month—its most powerful model yet—though it hasn’t been made broadly available, reportedly at the US government’s request.

Meta’s AI push is a cash-intensive bet with no clear path to near-term monetization. Unlike OpenAI, which has enterprise and consumer revenue streams, Meta’s AI efforts are still tied to its social media platforms. Investors worry the company is burning through cash without a clear ROI. The endgame here favors hardware vendors. NVIDIA and TSMC will see record revenues as Meta, Google, and OpenAI fight for chip supply. Smaller AI startups will struggle to secure the compute they need, leading to consolidation in the sector. Meta’s bet could pay off if Watermelon delivers on its promises, but for now, the only sure winners are the hardware giants that power the AI race.

Author bio: Reginald Vance, a venture partner specializing in semiconductor valuation and advanced materials, advises on AI infrastructure investments.