$META Muse is legit impressive — you can actually hear the infrastructure grinding in the background. All those $AMD, $INTC, and $ARM CPUs spinning up, plus the memory stack (NICs, SSDs, DRAM) needed to handle multiple browser tabs and VMs simultaneously.
When you throw long-form tasks at it, the resource demand becomes obvious. Intel's CEO recently mentioned CPU demand is so strong they can only meet 50% of customer orders — that's a real signal about infrastructure bottlenecks across AI deployments.
Also wondering how much $AMZN cloud or $NBIS components are involved in running Muse, or if Meta is handling everything in-house. That matters for understanding the broader capex cycle and who's winning the picks-and-shovels game.
Meta crushed expectations with this release. Still testing it out, but the infrastructure implications are bigger than people realize.
When you throw long-form tasks at it, the resource demand becomes obvious. Intel's CEO recently mentioned CPU demand is so strong they can only meet 50% of customer orders — that's a real signal about infrastructure bottlenecks across AI deployments.
Also wondering how much $AMZN cloud or $NBIS components are involved in running Muse, or if Meta is handling everything in-house. That matters for understanding the broader capex cycle and who's winning the picks-and-shovels game.
Meta crushed expectations with this release. Still testing it out, but the infrastructure implications are bigger than people realize.
