When OpenAI and Anthropic spent the week “pacing the frontier,” they probably didn’t expect Meta to walk away with the news cycle. Anthropic shipped Opus 5.5, and OpenAI answered with GPT-6 model updates just 90 minutes later, but the story that stuck, according to the TechCrunch Equity podcast hosts Kirsten Korosec, Anthony Ha, and Sean O’Kane, is Meta’s Muse. The personal AI agent is reportedly outpacing ChatGPT’s early mobile launch numbers, and Meta has plans that reach far beyond an app icon.
The surprise frontrunner
Muse’s early traction is the first signal that Meta’s consumer AI bet is working. Rather than trying to win a benchmark war, Meta is positioning Muse as an ambient companion: it’s already headed for smart glasses and a small, Tamagotchi-style wearable that TechCrunch has separately confirmed. The message is clear: this agent is meant to live on your face or in your pocket, not in a tab you remember to open.
At the technical core is Muse Spark, the first model out of Meta Superintelligence Labs. It has already replaced Llama 4 on most of Meta’s Ray-Ban and Oakley glasses. Muse Spark is a small, fast model that Meta claims matches Llama 4 Maverick with roughly 10x less compute, and it posts a 52 on the Artificial Analysis Intelligence Index. That’s behind Gemini 3.1 Pro, GPT-5.5, and Claude Opus 4.8, but for a wearable, raw leaderboard dominance may matter less than speed and efficiency. Unlike Llama, Muse Spark is not open source, a notable shift in Meta’s AI posture.
Why hardware changes the AI equation
A consumer agent with a hardware home competes on a different axis than a chatbot in a browser. Meta is betting that distribution through glasses and a pocket companion turns an AI assistant into a daily habit. The Tamagotchi-style device is especially telling: it frames Muse as a presence you carry and tend to, not a task you launch. For startups building on top of these agents, the question is whether Meta’s ambient approach can create the kind of switching costs that raw model quality rarely does.
Money moves behind the scramble
The same Equity episode touched on three very different funding stories that show where capital is flowing as AI becomes a platform play:
- a16z’s Horowitz Andreessen Academy is a new school for aspiring founders, which the hosts compare to the Thiel Fellowship and Silicon Valley’s long-running anti-college streak.
- Oura’s $2.2 billion IPO looks more like a payday for existing shareholders than a major capital raise for the company itself.
- Ema’s $77 million raise funds AI agent teams that automate HR, IT, and finance workflows, with Google and Microsoft already listed as early enterprise customers.
These aren’t all consumer plays, but they share a throughline: the scramble to own the interface where AI habits form. Meta is betting that interface is a face-worn or pocket-sized companion. This week, that bet pulled attention away from two frontier model drops—and that may be the more important signal.