14.8%. That's how much Meta Platforms' shares rose over the week, their largest weekly gain since at least February 2024. The spike followed three product moves: Muse Image, an updated Muse Spark 1.1 model aimed at agentic and coding tasks, and a new public developer API with explicit low-cost pricing of $1.25 per million input tokens and $4.25 per million output tokens. MarketWatch reported the stock jumped 6% on Friday, citing Dow Jones Market Data, a late-week pop that pushed Meta back into positive territory for the year and revived hopes that the company can translate heavy infrastructure spending into outside revenue.
6%. That was Friday's one-session gain, according to Dow Jones Market Data as reported by MarketWatch, and it drove a weeklong rally investors described as Meta's best showing in more than a year. Traders said the market treated the product rollouts and the new pricing as a clearer path to monetizing the company's AI work, not just its ad business.
Investors focused on two linked ideas. First, Meta showed technical progress with new models. Company leaders and executives at Meta Superintelligence Labs highlighted Muse Spark 1.1's capabilities, and Meta's chief AI officer Alexandr Wang signalled further iterations are planned. Second, Meta put explicit price points on access to its models, and CEO Mark Zuckerberg described the offering as available at a "very low price." The published API fees, $1.25 per million input tokens and $4.25 per million output tokens, were contrasted in market coverage with materially higher rates from some rivals, feeding hopes that Meta could undercut competitors on cost and accelerate usage.
The financial backdrop made those hopes powerful. Meta's first-quarter results in April prompted debate about the company's large infrastructure plan after management raised 2026 capital-expenditure guidance.
The company put 2026 spending in a range that analysts report as $125 billion to $145 billion, a scale that made monetizing spare capacity a strategic priority rather than a side project.
Some analysts now see a credible route to do that. Deutsche Bank analyst Benjamin Black wrote that reported capacity additions imply as much as $24 billion in potential incremental third-party cloud revenue, a figure notably higher than earlier estimates. Bank of America analysts flagged potential cost savings from Meta's plans for custom chips and data centres, a combination that could improve margins if execution holds.
Execution is the obvious caveat. Recent reports said Meta is on track to deploy roughly 7 gigawatts of compute this year and scale to about 14 gigawatts by 2027. Separate reporting also said Meta's custom AI chip, code-named Iris, cleared internal testing and is set to move into production in September. Broadcom is named as a design partner and Taiwan Semiconductor Manufacturing Co. will handle fabrication. Those two dates and technical steps are tangible milestones, but they also highlight the risks: chip production, unit compute costs and stiff competition from established cloud providers and model vendors.
The market reaction was concentrated among institutional investors and retail sentiment channels. Weekly gains erased year-to-date losses and narrowed the gap between Meta and other so-called Magnificent Seven stocks, though broad market indices still outpaced Meta's 2026 return. Coverage noted heightened bullishness on trading desks and retail forums, and a sustained acceleration in Meta's capex cadence would ripple to bond markets and suppliers tied to data-centre builds.
Related Articles
Meta's next public checkpoint is the second-quarter earnings report, where management could update capex guidance and monetization plans, while the planned start of Iris production in September will be the operational test of whether the company can materially lower its compute costs. Originally reported by marketwatch.com.
This article was created with AI assistance.
