Wall Street Interprets Meta (META.US) Latest Developments: Model Delays Highlight Google (GOOGL.US) Advantages; Cost-Cutting Through Layoffs Fall Short Against High AI Infrastructure Costs

Recently, news about Meta (META.US) delaying the release of its cutting-edge AI models and plans for layoffs has attracted market attention. Analysts generally expressed slight disappointment over the delays and believe that the cost-cutting layoffs will have minimal impact on its $169 billion spending plan through 2026.

Last Friday, reports indicated that Meta’s internal testing showed its AI model “Avocado” lagging behind competitors Google (GOOGL.US), OpenAI, and Anthropic in reasoning, coding, and writing capabilities, leading to the postponement of its release schedule.

According to reports over the weekend, to cope with high costs of AI infrastructure and achieve efficiency gains from AI, Meta is planning to cut more than 20% of its workforce.

AI Model Delays Draw Attention

JPMorgan emphasized that, aside from advertising, Meta’s AI models are another key component of its “bullish narrative,” so the delay was “somewhat surprising.” The firm pointed out that given Meta’s massive investment scale, its tolerance for failure is already quite limited.

JPMorgan analysts stated: “Frontier models are crucial for Meta to realize its superintelligence vision, maintain long-term control over its own computing platforms, and expand AI products beyond advertising. The market is also closely watching whether its substantial capital and operational expenditures related to the foundational model team will deliver expected results.”

Bank of America believes that, although the delay is “disappointing,” the revised timeline indicates that Meta is more focused on product performance quality rather than rushing to market.

Bank of America analyzed: “Considering that the foundational model team was only established at the end of Q2 2025 or early Q3, we think the original Q1 release target was quite challenging. The adjusted schedule instead reflects a more prudent development cycle. However, the challenges Meta faces also highlight Google’s strong position in large language models. Based on Gemini’s development trajectory, it may take several years for Meta to build a top-tier large language model.”

Impact of Layoffs

JPMorgan estimates that a 20% layoff could save Meta about $6 billion. However, even if this savings is fully converted into profit, it would have little “significant impact” on the company’s total expenditure forecast for this year.

The firm further estimates that if this $6 billion cost saving is included in 2027 profits and considering tax effects, it could increase GAAP earnings per share by about $2, higher than the current forecast of $31.50.

Bank of America shares a similar view, believing that cost reductions from layoffs are unlikely to materially lower the company’s annual expense guidance.

BofA estimates that this move could save the social media and tech giant up to $8 billion, but this only offsets part of the expected $45 billion GAAP expenditure growth in its models.

Notably, Jefferies analysts stated that if Meta is willing to make such large-scale layoffs while increasing AI investment, it indicates that “AI is increasingly becoming a key driver of productivity improvements.”

Jefferies further pointed out: “This is not only significant for Meta but will also profoundly impact the entire internet and software industry, prompting investors to reassess the relationship between company staffing, growth, and profit margins. We believe this could further pressure the SaaS sector’s seat-based charging model. Clearly, part of Meta’s layoffs are aimed at managing AI infrastructure costs, especially the sharp rise in capital expenditures.”

The firm added that, amid increasing macroeconomic uncertainty, Meta’s layoffs seem logical. Given the close correlation between advertising spending and GDP growth, the company is proactively optimizing efficiency to prepare for potential economic downturns, which is “entirely reasonable.”