Google's March Gemini limits pushed Meta toward Muse Spark update
So, in March 2026, Google basically told Meta it couldn't use as much of its Gemini AI as before because the demand was just too high.
This threw off a bunch of Meta's internal workflows and meant employees had to cut back on using AI tokens, their way of tracking system activity.
Weeks before this, Alexandr Wang announced that a new Muse Spark update was coming soon, Meta's in-house model that's now stepping up for tasks Gemini used to handle.
Meta commits $600B, plans $145B AI
Meta originally picked Google's Gemini over its own Llama models because Gemini was better at things like scam detection and content moderation.
But with these new restrictions, they're putting more energy into Muse Spark, which launched in April and is seen internally as a stronger rival.
Plus, Meta isn't holding back: they've committed $600 billion for US investments by 2028 and are planning up to $145 billion just this year on AI infrastructure.