Are we funding a hollow AI bubble?
Ed Zitron points out the contradiction: AI providers aren't charging for outcomes, they’re charging for wasteful token consumption.
They don't know what you can build, so they force you to foot the bill for their experimentation.
Is it time to move past this model and shift toward a more distributed, wider ownership of compute?
By democratizing access to infrastructure, we move away from rent-seeking gatekeepers and toward a system where value, not just token usage, defines the ROI.
The writing is on the wall.
Public support for hyperscale data centers is collapsing, and 77% of Americans are now worried AI will drive up their electricity bills, and only 14% are okay with a data center near their home.
NY just became the first state to enact a statewide moratorium, and 300+ data center bills have been filed across 30+ states, and $64B in projects have been blocked or delayed.
Nvidia sees it too. They're pivoting to distributed compute: residential nodes, personal AI supercomputers, and a vision of "an AI supercomputer in every household."
The narrative is shifting from scale to sovereignty.
The question is no longer how big your cluster is, it's who owns the compute.
"The jig is up."
Palantir cofounder Alex Karp on what technical customers actually want:
"control over their compute, their models, their data stack, and their alpha."
Blind reliance on centralized providers is ending. Replaced by permissionless compute and true ownership.