Meta Platforms, Inc. (NASDAQ: META) shares surged this week, driven by a series of strategic advancements in artificial intelligence, marking its strongest weekly performance since February 2024 and signaling a significant recovery in market confidence regarding its AI capabilities and commercialization path.
On Friday alone, the stock climbed 6%, culminating in a weekly gain of 14.8%. This rally, the most robust since at least February, has pushed the company's year-to-date performance into positive territory, up approximately 1.4%.
Catalysts for the Surge
The last time Meta experienced a comparable weekly surge was in February 2024, when investors reacted positively to early results from its "Year of Efficiency" cost-cutting initiative, designed to repair the financial image tarnished by heavy bets on the metaverse. The recent price breakthrough suggests the company is shedding its market perception as an "AI laggard," potentially creating more room to intensify its AI strategy.
A key development was the July 9th launch of Meta's flagship model, Muse Spark 1.1. Reports indicate it outperformed Alphabet's (NASDAQ: GOOG) Gemini model in several benchmark tests, including agent capabilities, coding, and multimodal tasks.
Concurrently, reports suggest Meta is advancing plans for mass production of its in-house designed chips and significantly expanding its computing infrastructure. This prompted Deutsche Bank analyst Benjamin Black to raise his estimate for the company's potential incremental revenue from third-party cloud services from $17 billion to $24 billion.
Research firm SemiAnalysis published a report forecasting that Meta's AI research division, Meta Superintelligence (MSL), could surpass Alphabet in cutting-edge AI capabilities within the next six months. This could shift the competitive landscape from a duopoly of Alphabet and OpenAI to a three-way race involving Meta, OpenAI, and Anthropic.
Aggressive Pricing Strategy
The Muse Spark 1.1 model, launched this week, is Meta's first commercially available model with near-state-of-the-art agent programming capabilities, featuring a paid API. CEO Mark Zuckerberg emphasized on social media platform X that the model's pricing is "very low-cost," sparking widespread market speculation that Meta is initiating an AI inference pricing war to pressure competitors.
The Meta Model API offers a $20 free credit per account, with a pay-as-you-go model charging $1.25 per million tokens for input and $4.25 per million tokens for output. Richard Windsor, founder of Radio Free Mobile, noted in a Friday report that the launch of Muse Spark 1.1 aligns with recent reports about Meta planning a new business selling computing power.
Windsor wrote that increasing evidence suggests Meta will launch a new business selling computing power to third parties, given the currently attractive returns. He further pointed out that Muse Spark's AI programming capabilities are close to top-tier models, "but priced at just 25% of their cost," making it highly attractive to the mass market.
In-House Chips and Computing Expansion
Reports indicate Meta plans to begin mass production of its in-house AI chip, codenamed "Iris," in September. The chip, co-designed with Broadcom and manufactured by TSMC, completed testing in just six weeks. Meta has also signed multi-year supply agreements with Samsung, SanDisk, and Sumitomo Electric.
In terms of computing scale, Meta aims to deploy 7 gigawatts of computing power this year, doubling that figure to 14 gigawatts by 2027. Supporting this goal is the simultaneous construction of five gigawatt-scale "titan" hyperscale data center clusters and a proprietary "AI-Backbone" network architecture, which allows Meta to asynchronously scale complex training tasks across thousands of kilometers.
Deutsche Bank's Benjamin Black stated in a Thursday report that this computing expansion plan implies potential incremental revenue from third-party cloud services of approximately $24 billion for Meta, significantly higher than the previous forecast of $17 billion. He also noted that the in-house chips could open a viable path for the company to reduce costs and improve efficiency.
Analyst Outlook on Competitive Landscape
The SemiAnalysis report posits that after a year of aggressive capital investment and organizational restructuring, MSL could surpass Alphabet in frontier AI capability rankings within six months. The report suggests the existing duopoly of Alphabet and OpenAI could be rewritten into a three-way contest between Meta, OpenAI, and Anthropic.
The core judgment hinges on the speed of computing power expansion. Meta's growth trajectory in AI computing scale is projected to exceed the combined computing power of OpenAI and Anthropic by year-end. Reports cite an internal memo indicating Meta's capital expenditure ceiling for AI infrastructure this year is as high as $145 billion.
On the talent front, Meta has reportedly reassigned 3,000 engineers to an internal reinforcement learning environment factory to build proprietary data difficult for commercial data vendors to replicate. It has also invested $14.3 billion in Scale AI, facilitating the large-scale recruitment of top researchers from organizations like OpenAI and Anthropic.
SemiAnalysis argues that evaluating MSL based solely on current benchmark scores is "missing the forest for the trees." The crucial factor is future momentum, not the current starting point. The report concludes that if Zuckerberg maintains the current pace of capital investment, Alphabet risks being permanently excluded from the top tier of global AI hyperscalers.
Comments