Mark Zuckerberg said AI was behind schedule last week but Meta's up by 22% in last 10 days.

Meta was up about 6% today and gaining roughly 22% over the last 10 trading days. That's a genuinely interesting move because earlier this week Zuckerberg himself said the company's AI agent progress had been 'slower than expected', and before that the stock had been under real pressure over capex fears, $145 billion in planned 2026 AI infrastructure spending with investors nervous about a clear payoff. So, despite Zuck admitting that their work on AI was behind schedule the market thought the other way.

Few things to note, Bank of America maintained its buy rating and pointed to an internal memo, (reviewed by Reuters), suggesting Meta is meaningfully improving its AI cost structure, which directly addresses the exact concern that had been weighing the stock down. Then Meta launched Muse Spark 1.1 explicitly calling it their strongest model yet for agentic and coding work and positioning it as competitive against Anthropic and OpenAI. And earlier this week there was also the Iris chip news, Meta's in-house AI chip hitting a real production milestone after a previously troubled development history.

So in the space of about a week, Meta answered three separate bear arguments almost one after another: the cost structure concern, the capability of AI concern, and the dependency on chipmakers concern. That's a lot of narrative-shifting news happening in a short window and the market's reaction suggests investors are reading it as a sign of real progress rather than three unrelated announcements.

On the bearish side, going from being skeptical about own's AI progress to admitting that the AI spending is improving structurally with an internal memo and launching one model update should not be the only reason for this positive move. It will be fair to wonder if this could be a well-timed run of good headlines landing right before earnings, which is when this actually gets tested with real numbers instead of memos and model releases.

So should a week like this change your view on Meta's AI execution specifically, or does this feel like good headline building before the earnings.