Ferrari proposes that the next constraint on AI lies in physical infrastructure

On October 19, 2026, Moby Chief AI Officer Michael Ferrari presented a judgment at Cypher 2026, arguing that the next constraint on AI growth comes from physical infrastructure such as energy, supply chains, and water resources, not just chips and models. Citing PJM as an example, he said grid-connection waits can reach 5 to 7 years, available power headroom has fallen from about 70GW in 2022 to 18GW, and a gap of about 40GW may emerge within five years. He also proposed undersea and orbital computing as alternative directions, but estimated that current space computing costs about 4 times per GPU-hour that of ground-based computing, and that large-scale commercialization may not come until the 2030s or even 2040. Currently this judgment is reported by a single source, with no subsequent factual developments.

AI-generated from public reports

1. Oct 19· 1 reports

   Ferrari proposes that AI's next constraint lies in physical infrastructure

   Source:Analytics India Magazine