Meta’s latest AI breakthrough, the Muse Spark model, marks a pivotal moment in the company’s effort to regain ground in the high-stakes race against OpenAI, Google, and Anthropic—but it is also exposing deep internal disagreement over how far, and how fast, the company has really come.

A year of “wartime” AI at Meta

About a year ago, Mark Zuckerberg handed control of Meta’s AI reboot to Alexandr Wang, then a 28‑year‑old start-up founder, in a bid to “jolt” the company’s artificial intelligence efforts into “wartime mode.”¹ The choice of an outsider over veteran researchers signaled a desire for urgency and radical change.

Over the following 12 months, Wang built an elite research unit, known as TBD Lab, hiring top talent on multimillion‑dollar salaries and reshaping parts of Meta’s AI organization.¹ His rising influence was underscored when he became the only Meta leader besides Zuckerberg to attend a White House dinner with top Silicon Valley figures hosted by President Donald Trump.¹

Muse Spark arrives—and divides opinion

In April, TBD Lab released Muse Spark, described as “Meta’s most credible AI model yet” and the first major system to emerge from Wang’s secretive team.¹ Supporters inside and outside the company say the model has “brought momentum” to Meta’s AI efforts and see it as “the clearest sign yet that Meta’s AI rebuilding effort is gaining traction.”¹ Some experts argue that successor models expected in the coming months could further close the gap with leading rivals.²

Enthusiasm vs. skepticism

Proponents praise Wang’s pace and pragmatism. One former Meta AI research chief said “the amount of work the TBD Lab was able to do in a short amount of time is very impressive,” adding that “Alex knows what he doesn’t know and he’s willing to listen.”¹

Yet critics inside Meta describe his leadership as frenetic and argue that he has oversold what they see as incremental gains, contending that “doubts linger over whether Meta can close the gap with rivals” and achieve a leading position in frontier AI.²

As Muse Spark rolls out and new models loom, the core tension endures: has Meta truly reignited its AI edge, or is it still playing catch‑up under a high‑profile, high‑pressure experiment in outsider leadership?