Meta is betting its AI future on Muse, a personal agent less than one month old, while planning to spend up to $145 billion on infrastructure this year. At its developer conference in Menlo Park, Zuckerberg called Muse 'the centerpiece of our vision' and projected it will become 'the personal superintelligence that billions of people around the world are going to use.' The numbers so far do not support that scale: week-one users topped 500,000 per The Information, while Meta's own Threads hit 100 million users in five days. The company is already building hardware around Muse, including the Muse Charm, a strapless smartwatch with an animated agent interface scheduled to ship this year at an undisclosed price.
The article is worth reading in full not for the Muse product breakdown, but for the historical pattern it traces. The author was at Meta's 2016 bot platform launch, watched it collapse by 2018, and is now watching the same developer-conference playbook repeat. The core critique is specific: Muse behaves more like an intern than a coworker, requiring constant task assignment, action approvals, and output review, which maps it closer to enterprise software than to a mass-market consumer product. Meanwhile, Meta's Reality Labs has lost over $85 billion since 2021, and this year's capital spending alone is set to exceed those cumulative losses.
The piece does not let Zuckerberg off the hook for the metaverse pivot. He acknowledged at the keynote that he misjudged VR's timeline relative to AI, but the author documents how Muse is now occupying the exact rhetorical slot the metaverse held at prior conferences: a rally point deployed while core-product litigation and social backlash go unmentioned. The one genuine bright spot the author credits is Meta's VR Glasses, a separate announcement. The question the full article forces is whether momentum and capital can substitute for product-market fit when the previous billion-dollar answer to that question is still on the balance sheet.
[READ ORIGINAL →]