Executive Summary
Anj Midha reveals a critical infrastructure crisis hiding beneath AI's exponential growth: billions of dollars in stranded GPU compute sitting unutilized due to lack of standardization. Unlike electricity's transition from private generators to the grid in 1885, AI compute remains fragmented across incompatible chip generations and proprietary systems. Midha's AMP is building the 'compute grid' - an independent system operator coordinating 1.3 gigawatts of capacity across frontier labs. This addresses what he calls the 'GPU wastage bubble' where H100 clusters become obsolete for frontier training but can't efficiently serve inference workloads. The standardization crisis extends beyond technical compatibility to geopolitical vulnerability, as Chinese state-sponsored distillation attacks exploit Western infrastructure fragmentation. Midha's thesis: we're not in an AI capabilities bubble but an infrastructure coordination crisis that threatens frontier progress. His solution combines compute pooling, adversarial defense coordination, and mission-aligned capital allocation. The window for Western AI sovereignty closes as China executes full-stack systems co-design while the West struggles with misaligned incentives between hyperscalers, labs, and government procurement standards.
Key Insights
what Anj Midha said“We are definitely in a GPU wastage bubble, where there are stranded pockets of compute, like billions of dollars of compute, that are sitting unutilized. If we could pull them together on a grid across the independent ecosystem.”
This is a preview. Log in to see the full analysis including investment opportunities, risks, catalysts, and detailed insights.