Meta is spending up to $145B on AI this year, and it’s changing how the company builds everything

Meta plans to spend between $125B and $145B on AI infrastructure in 2026, with CEO Mark Zuckerberg confirming that AI tools now generate most of the company's code and are accelerating product development across Facebook, Instagram, and Marketplace. The company aims for select engineering teams to have over 75% of committed code produced by AI by mid-2026, and it released the Muse Spark 1.1 multimodal reasoning model via a public API on July 9.

Via browseract.com Meta is spending up to $145B on AI this year, and it’s changing how the company builds everything Mark Zuckerberg says AI tools now generate most of the code at Meta, with implications that ripple well beyond Menlo Park Meta just told investors something that should make every developer, founder, and crypto builder pay attention: AI has fundamentally changed how fast the company ships products. CEO Mark Zuckerberg confirmed during Q2 2026 earnings commentary that AI is accelerating development across Facebook, Instagram, and Marketplace, with a wave of new consumer products already in the pipeline. The numbers backing that claim are staggering. Meta’s 2026 capital expenditure guidance sits between $125B and $145B, nearly all of it funneled into AI infrastructure. The company also plans to double its cloud capacity this year. AI is writing the code now Meta is targeting a milestone where select engineering teams generate over 75% of their committed code using AI tools by mid-2026. Zuckerberg indicated the company is making significant progress toward that goal. The company released Muse Spark 1.1 on July 9, a multimodal reasoning model designed for agentic tasks and coding. It’s available through a new public Meta Model API, which means outside developers can tap into the same kind of tooling Meta uses internally. Why crypto builders should care Meta didn’t mention crypto, blockchain, or Web3 in any of its recent announcements. Not once. But the downstream effects of what Zuckerberg is describing matter enormously for the decentralized app ecosystem. The single biggest bottleneck for Web3 consumer products has never been the blockchain layer. It’s been the application layer. Building polished, user-friendly apps on top of decentralized infrastructure is expensive, slow, and requires specialized talent. If AI tools can compress development timelines by 50% or more, that constraint loosens for everyone, not just Meta. The Muse Spark 1.1 model being publicly accessible via API is particularly relevant. Crypto projects that need multimodal reasoning capabilities, think AI agents that can parse images, text, and code simultaneously, now have another tool in the stack. The infrastructure arms race and its market implications Meta spending up to $145B on capex in a single year is not normal. To put that in perspective, it’s roughly equivalent to the entire GDP of Hungary. The company is betting that AI infrastructure will be the defining competitive moat of the next decade, and it’s willing to compress its margins in the near term to get there. The doubling of cloud capacity also matters for the broader compute market. When Meta absorbs that much GPU and data center capacity, it tightens supply for everyone else. Decentralized compute networks that can aggregate idle resources become more attractive when centralized cloud providers are constrained. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy https://cryptobriefing.com/editorial-policy/ .