The world runs on electricity, but nobody builds their own power plant. Anjney Midha thinks AI compute should work the same way. His company, AMP PBC, is building what amounts to an electric grid for GPUs, pooling underutilized chips from independent labs and data centers into a shared network that member AI labs can tap on demand.
It’s a simple idea with a complicated execution. And Midha has $1.3 billion in funding commitments from backers including Andreessen Horowitz and Y Combinator to figure it out.
The GPU crunch is real, and getting worse
Here’s the thing about GPUs right now: there aren’t enough to go around, and the ones available aren’t cheap. Cloud GPU rental prices have surged, with hourly rates for B200 units reaching $4.89. Those rates have roughly doubled since January 2026, squeezing smaller AI teams that can’t match the purchasing power of Big Tech.
This is the bottleneck Midha identified after leaving his role at Andreessen Horowitz. The venture capital world gave him a front-row seat to the problem. Startups with promising AI research kept hitting the same wall: they couldn’t get enough compute at a price that made sense. Meanwhile, data centers and independent operators were sitting on GPUs that weren’t running at full capacity.
