The artificial intelligence arms race has turned Silicon Valley into a spending contest. Meta Platforms (NASDAQ:META | META Price Prediction), Microsoft (NASDAQ:MSFT), Alphabet (NASDAQ:GOOG), and Amazon (NASDAQ:AMZN) are pouring hundreds of billions of dollars into chips, servers, networking gear, and data centers in a scramble to dominate AI infrastructure. The question investors are wrestling with is simple: how much spending is too much?
For Meta shareholders, that concern just intensified. CEO Mark Zuckerberg told investors at Meta’s annual shareholder meeting on Wednesday that the company expects capital expenditures to reach as much as $145 billion in 2026. That figure dwarfs the company’s already massive spending plans and cements Meta’s transformation from a social media company into a full-scale AI hyperscaler.
But surprisingly, Zuckerberg may have also revealed the safety valve if Meta builds too much capacity.
Meta’s AI Ambitions Have Overtaken the Metaverse
Just three years ago, Meta made itself synonymous with the metaverse. The company spent tens of billions of dollars through its Reality Labs division chasing virtual reality adoption that never fully materialized, generating a cumulative operating loss exceeding $70 billion between 2021 and 2025.
Now the focus has shifted almost entirely to AI. Unlike the metaverse push, however, Meta’s AI spending is tied directly to its core business. AI recommendation engines already power content discovery across Facebook, Instagram, and Threads. Advertising tools driven by generative AI have improved campaign targeting and boosted engagement metrics that feed Meta’s advertising machine — still responsible for roughly 97% of revenue.
Let’s put the scale of this spending into perspective:
| Company | 2026 Estimated Capex | Primary Focus |
| Meta Platforms | $125 billion to $145 billion | Internal AI infrastructure |
| Microsoft | Roughly $190 billion | Azure + OpenAI infrastructure |
| Alphabet | $175 billion to $190 billion | Google Cloud + Gemini AI |
| Amazon | Roughly $200 billion | AWS expansion |
The key difference is that Meta is not primarily building cloud infrastructure for outside customers. At least not yet.
That distinction matters because investors fear Meta could repeat the same pattern seen during the metaverse era — huge spending without a clear return on investment.
Zuckerberg’s Backup Plan Changes the Equation
That concern helps explain why Zuckerberg’s comments at the shareholder meeting landed differently than past spending announcements.
Zuckerberg said launching a cloud computing business is “definitely on the table” if Meta ends up building excess AI capacity. He noted that third parties regularly approach Meta seeking access to its computing power and are willing to pay premium pricing for it.
In short, Meta may be building an escape hatch into one of tech’s most profitable businesses. That would place Meta directly into competition with:
- Amazon Web Services
- Google Cloud
- Microsoft Azure
Granted, those businesses took years to scale. Amazon Web Services generated $128.7 billion in revenue last year, while Microsoft’s Intelligent Cloud division surpassed $131 billion in annualized revenue. Still, Meta has a history of entering proven markets late and succeeding anyway.
Instagram Stories copied Snapchat. Reels copied TikTok. Threads emerged after turmoil at X. Meta rarely invents the category — but it often wins through scale, engineering talent, and distribution. A cloud infrastructure business would follow the same playbook.
Investors Have Good Reason to Be Nervous — But There’s Also Opportunity
That said, skepticism is understandable. Meta stock lost more than 75% between late 2021 and late 2022 as investors questioned Zuckerberg’s metaverse spending spree. The market has not forgotten that lesson.
At current levels, Meta trades at a premium valuation compared to its historical average because investors expect AI investments to produce real earnings growth. If spending accelerates faster than revenue, margins could narrow and pressure the stock.
Regardless, Meta’s position is stronger today than during the metaverse buildout. The company generated more than $45 billion in free cash flow over the past four quarters, giving it substantial financial firepower to fund AI expansion without crippling its balance sheet.
Zuckerberg may be making a calculated bet that AI infrastructure itself becomes the next platform business — much like cloud computing did a decade ago.
Key Takeaway
Meta’s $145 billion AI gamble carries real risk. Sharp investors should not ignore the possibility that spending outruns returns, particularly after the company’s costly metaverse detour.
But Zuckerberg’s comments revealed something investors may have overlooked: Meta is not simply building AI infrastructure for experimentation. It may be building the foundation for an entirely new business line.
If Meta eventually monetizes excess capacity through cloud services, today’s spending could look less like reckless expansion and more like the early stages of another tech empire. For long-term shareholders, that possibility changes the conversation entirely.