AMP PBC's Goal
Anjney Midha's new project, AMP PBC, seeks to radically lower the price of compute by building a grid that turns GPUs into a standardized utility.
56Odd Lots
Anjney Midha's Plan to Radically Lower the Price of Compute
Former a16z partner Anjney Midha explains how AMP PBC aims to reduce AI compute costs by standardizing GPUs and improving utilization.
Listen to the source episode
Anjney Midha's Plan to Radically Lower the Price of Compute
Continue in GenPod
Turn this episode into your next question.
Ask naturally. GenPod can find a related episode—or make a focused explainer when the right one does not exist.Key takeaways
What to know before you press play.
Current Market Fragmentation
Midha argues that compute is currently too fragmented, forcing labs to spend money on capacity that often goes unused due to contract structures.
78Challenges for Small Labs
Small labs are forced to pay for large, long-term contracts despite having spiky demand during model training, leading to inefficiencies.
9Software Solution for Utilization
Midha believes a software solution can significantly improve compute utilization, addressing the inefficiencies in the current market.
10Diverse AI Model Landscape
Midha anticipates that no single company will dominate; instead, a wide range of models will emerge, each optimally used in specific applications.
11Ask GenPod next
Keep the question moving.
“How does AMP PBC's grid technology differ from existing cloud providers?”
“What specific software solutions are being developed to improve GPU utilization?”
“How might the standardization of GPUs impact small AI research labs?”
Background reading
The context behind the episode.
Questions
Questions to carry into the episode.
What is the main problem with current compute contracts?
Labs are forced to spend money on capacity that often goes unused, particularly small labs paying for big, long-term contracts with spiky demand.
89How does AMP PBC plan to solve this?
By building a compute grid that turns GPUs into a standardized utility to improve utilization.
610What does Midha predict about the future of AI models?
He predicts a wide range of models will emerge, each optimally used in specific applications, rather than one dominant player.
11Experts
Voices named in the source.
Sources and disclosure
Where this guide comes from.
This guide is based on the episode 'Anjney Midha's Plan to Radically Lower the Price of Compute' from the podcast Odd Lots, published on June 13, 2026.
- Publisher show notes
Anjney Midha wrote the first check to Anthropic.
- Publisher show notes
He teaches a viral course at Stanford on how AI works.
- Publisher show notes
And he was, until recently, a partner at a16z.
- Publisher show notes
In other words, he is AI-industry royalty.
- Publisher show notes
Midha's new project is AMP PBC, a company that believes it can radically lower the price of compute.
- Publisher show notes
To accomplish that, he is working on building a compute grid that turns GPUs into a standardized utility.
- Publisher show notes
But right now, compute is too fragmented.
- Publisher show notes
And given the way contracts are structured, he says that labs are being forced to spend money on capacity that often goes unused.
- Publisher show notes
In other words, small labs are forced to pay up for big, long-term contracts, even though their own demand (particularly during model training) may be very spiky.
- Publisher show notes
On this episode, Midha explains how the market for compute currently works and why he believes there's a software solution that could significantly improve compute utilization.
- Publisher show notes
He also tells us why he does not anticipate one company will emerge as the dominate player and that instead we'll have a wide range of models, each optimally used in specific applications.
- Publisher show notes
Read more: Amazon Says Its Data Centers Use 2.5 Billion Gallons of Water Oracle Falls Most in Six Months on Mounting Data Center Costs
- Publisher show notes
Only http://Bloomberg.com subscribers can get the Odd Lots newsletter in their inbox each week, plus unlimited access to the site and app. Subscribe at bloomberg.com/subscriptions/oddlots
- Publisher show notes
Subscribe to the Odd Lots Newsletter Join the conversation:discord.gg/oddlots
- Publisher show notes
See omnystudio.com/listener for privacy information.
Related listening guides