Key Takeaways
- Shares of Meta climbed 4% on September 3 following the launch of Muse Spark 1.3, an enhanced AI coding platform
- Bank of America maintained its Buy recommendation with an $810 price objective, suggesting 32% potential gains from September 3 closing levels
- The latest Muse Spark iteration delivers 20% reduction in tool calls and 25% fewer tokens versus its predecessor
- Custom MTIA silicon developed with Broadcom partnership may account for 15-20% of Meta’s AI infrastructure
- The social media giant trades at 18x forward 2027 GAAP earnings, below its historical valuation average
Shares of Meta Platforms closed at $613.62 on September 3, registering approximately 4% gains. The broader equity market advanced roughly 1%, indicating Meta-specific catalysts drove the outperformance.
The catalyst? Muse Spark version 1.3.
The company unveiled this upgraded model on September 2. Engineered for coding applications and agentic workflows, it’s accessible via Muse Code and the Meta Model API. The platform achieves comparable engineering outcomes using approximately 20% fewer tool calls and 25% reduced token consumption compared to Muse Spark 1.2.
The pricing structure remained unchanged at $1.25 per million input tokens and $4.25 per million output tokens. Meta is directly challenging Anthropic’s Claude Code and OpenAI’s development tools in the agentic coding arena.
The enhanced model delivers improved reliability for extended tasks, orchestrates multiple workflows within single conversations, and integrates contextual information from diverse sources.
What captured Wall Street’s attention was the velocity of innovation. Muse Spark 1.3 launched approximately 30 days after version 1.2. Such rapid iteration represents an aggressive development timeline.
Bank of America’s Bullish Perspective
Bank of America analyst Justin Post reaffirmed his Buy recommendation and $810 price objective following the model launch. This target represents approximately 32% upside potential from the September 3 closing price.
Post’s analysis highlighted the agentic enhancements as foundational infrastructure for Meta’s upcoming initiatives. The firm is working on a consumer-facing AI agent with the internal designation Hatch, initially disclosed by The Information. Neither an official brand name nor release timeline has been announced.
Bernstein similarly maintained an Outperform stance with an $800 objective, emphasizing Meta’s AI-powered advertising platform as a competitive differentiator.
Trading around $617 when BofA issued its analysis, Meta carried a valuation of approximately 18 times estimated 2027 GAAP earnings. The company’s historical average hovers near 21 times. The S&P 500 currently trades at roughly 20 times. Meta’s valuation sits below both its own historical norm and the benchmark index.
BofA’s $810 objective assumes 24 times 2027 GAAP earnings. The firm contends this premium valuation is warranted given Meta’s expansion trajectory.
Custom Silicon Initiative
The investment thesis extends beyond software developments.
During its Q2 2026 earnings discussion, Broadcom disclosed expectations to supply three successive generations of Meta’s proprietary Training and Inference Accelerator processors through 2027. The chipmaker also has visibility into approximately three gigawatts of Meta infrastructure deployments extending through 2028, with volume shipments anticipated to commence in Q4 2026.
BofA projects these MTIA implementations could ultimately comprise 15% to 20% of Meta’s aggregate AI computational capacity.
KeyBanc maintained an Overweight stance but reduced its price objective to $760 from $855. The research shop acknowledged Meta Superintelligence Labs achieved substantial advancement with Muse Spark while noting escalating expectations for demonstrable AI monetization.
Meta’s 52-week trading range spans from $520.26 to $790.80. Following the September 3 appreciation, shares remain positioned in the lower portion of this range.
Volume production of MTIA processors is slated to begin during Q4 2026.



