Meta shares rallied through the week to close up roughly 15%, marking the stock's best weekly performance since early 2024, as investor sentiment shifted decisively in favor of Mark Zuckerberg's AI strategy after a rapid-fire sequence of model launches and infrastructure disclosures. The rally capped a week that began with skepticism about Meta's AI capital spending and ended with the market treating that same spending as a credible, monetizable bet.
The catalyst was product cadence. Three months after introducing Muse Spark, its first proprietary foundation model, Meta shipped Muse Image on Tuesday -- a new AI image-generation model aimed squarely at creators and advertisers using Meta's subscription products -- and followed it two days later with Muse Spark 1.1, a model built specifically for agentic and coding workloads, entering the same crowded coding-model category OpenAI's GPT-5.6 and SpaceXAI's Grok 4.5 are fighting over. Two major model releases in the same week is an unusually fast cadence even by 2026 AI-industry standards.
The infrastructure story compounded the enthusiasm. Reports surfaced that Meta plans to begin production of its custom "Iris" AI chip in September 2026, a silicon effort expected to roughly double the company's internal computing capacity and reduce its dependence on Nvidia GPUs for at least a portion of its workloads. Meta is targeting a total of 7 gigawatts of computing infrastructure by the end of 2026, scaling to 14 gigawatts by 2027 -- a buildout pace that, until this week, investors had mostly treated as an open-ended cost center rather than a business with a visible return.
“Two major model releases in the same week is an unusually fast cadence even by 2026 AI-industry standards.”
What changed the market's read isn't just the chip or the models individually, but Meta finally articulating a coherent narrative connecting them: a proprietary model stack, custom silicon to run it more cheaply, and infrastructure scaled large enough that Meta could plausibly rent out excess capacity by competing in cloud computing against Amazon Web Services and Microsoft Azure. That's a materially different pitch than "we're spending tens of billions on AI because everyone else is," and it's the pitch that moved the stock.
For founders and operators watching Big Tech capital allocation as a leading indicator, Meta's rally is evidence that public markets will reward AI infrastructure spending once a company can show product output and a monetization thesis attached to it, not just capex growth in isolation. For competitors in AI coding tools specifically, Meta shipping Muse Spark 1.1 with real distribution behind it -- inside the Meta AI app and meta.ai -- adds a fourth credible, well-capitalized competitor to a category that startups increasingly can't win on model quality alone.
The bear case: a single strong week doesn't resolve the deeper question of whether Meta's AI capex, still scaling toward 14 gigawatts by 2027, generates returns commensurate with the spending, and any stumble in Iris chip production or Muse model adoption could reverse sentiment just as quickly as it built. What to watch next: independent benchmarks of Muse Spark 1.1 against GPT-5.6 and Grok 4.5, and whether Meta discloses more specifics on a potential cloud-computing business at its next earnings call.