Meta shares have surged 30% in September following the release of its Muse personal AI assistant, marking the stock's best month since July 2013 and positioning it on the cusp of joining an elite group of companies worth at least $2 trillion. The stock has been the third-best performer in the S&P 500 with a 43% gain since the recovery began after Meta agreed to pay up to $18 billion to settle a social-media lawsuit. This dramatic reversal comes after Meta shares were down 18% for the year less than six weeks ago, putting the stock among the 50 worst-performers in the S&P 500 through August 18. As per The Hindu BusinessLine, the recovery has been so significant that it lifted Meta shares nearly 12% in a single day, contributing to Mark Zuckerberg's estimated net worth jump of $25 billion as reported by Forbes. The sharpest move came on September 21, when shares jumped 11% to close at $741.25, marking Meta's biggest single-day gain since April 2025. On Thursday, Meta shares closed 4.5% higher at $777.59, with 57 of the 63 analysts covering Meta rating it a buy or strong buy according to LSEG data.
The Muse AI assistant, launched on September 8 in the United States and September 18 in Canada, has demonstrated exceptional early traction with 1.8 million downloads in less than two weeks according to Apptopia data, as reported by AFP. When compared to similar metrics for OpenAI's ChatGPT in its first 12 days, Muse achieved 1.8 million downloads versus 1.3 million for ChatGPT. The app has climbed to the top of the free-app rankings in both the United States and Canada, with daily US downloads now running ahead of Meta's other major applications including Facebook, Instagram and WhatsApp. J.P. Morgan analysts believe Muse has "the potential to become the most widely used consumer AI application since ChatGPT," crediting its early success to strong product-market fit, Meta's distribution reach, and a free tier that lowers barriers to trial and repeat use. Early reviews suggest "functionality is resonating with users, particularly the agent's ability to execute multi-step tasks and workflows with limited user intervention," according to BofA Global Research. J.P. Morgan reports that Muse has connected with more than 2,000 applications in two weeks and has major integrations with Walmart, Best Buy, Sephora, and Wayfair. The app had briefly topped Apple's US App Store chart for free apps, demonstrating strong consumer demand.
The launch of Muse represents a pivotal shift in AI infrastructure requirements, moving beyond traditional chatbot models to agentic AI that performs tasks rather than just answering questions. As reported by Investing.com India, Meta's Chief AI Officer Alexandr Wang stated that "early usage has blown way past our projections," with users consuming 10x more than Meta's testing groups. The app requires two virtual CPUs, memory and storage per user, with no GPU processing, fundamentally changing the hardware requirements for AI inference. This shift from 1 CPU for every 7-8 GPUs in training to 1 CPU for every 1-3 GPUs in inference means AI agents require approximately 1:1 CPU-to-GPU ratios, compared to the 1:10 ratio seen in traditional AI systems. The 2.8 million downloads in first 12 days across the US and Canada represent more than double ChatGPT's pace over the same period, with the app passing 500,000 users in its first week. Meta plans to profit over time by "taking a small fee from transactions" - the agent gets paid when it completes tasks, marking a new business model for AI monetization.
The agentic AI revolution is creating unprecedented demand for CPUs, with Investing.com India reporting that Intel's CEO says the company can meet only about half of customer demand for its CPUs. AMD's server CPU sales grew 70% last quarter, while Anthropic signed an $11.6 billion, seven-year deal with Akamai for CPU capacity. The market opportunity has expanded dramatically, with AMD's estimate for the 2030 server CPU market growing from $60 billion in November 2025 to $220 billion by July 2026. Arm projects that inference will overtake training as the largest AI workload in data centers by 2030, growing about 35% annually. The current production capacity of 35-45 million server CPUs annually would need to increase by more than 20x to meet projected demand from 1 billion knowledge workers running 10 agents each. Intel and AMD build x86 chips, while Arm designs architecture for custom chips, with Arm-based chips representing the fastest-growing share of the server market. AMD's Lisa Su noted that the ratio is moving "closer to 1:1," fundamentally changing the hardware mix for AI infrastructure.
Wall Street analysts are increasingly bullish on Meta's AI monetization prospects, with J.P. Morgan raising its price target to $920 from $820 on Thursday, maintaining an overweight rating. J.P. Morgan analyst Doug Anmuth wrote that "the agent is off to a strong start and expanding quickly," with the bank expecting Muse engagements to be handled through "agent-to-agent interactions" allowing Meta to monetize on a take-rate or commission model. Wells Fargo made a similar case, with analyst Ken Gawrelski raising his target to $796 from $640, citing early demand for Muse. Jefferies estimates that if Muse could reach 1 billion users by end of 2027 and at least 3% transition to paid versions, it could generate $10.8 billion in annualized revenue. Of the 64 brokerages that cover Meta's stock, 57 rate it "buy" or higher with a median price target of $760, according to data compiled by LSEG. The enthusiasm around Muse has sparked a broader Wall Street tech rally, with Meta shares lifting chipmakers including AMD and Intel, with AMD crossing the $1 trillion valuation mark for the first time.
