Five AI supply chain experts discuss chip, energy and architecture bottlenecks
Updated
Updated · TechCrunch · May 7
Five AI supply chain experts discuss chip, energy and architecture bottlenecks
10 articles · Updated · TechCrunch · May 7
At the Milken Global Conference in Beverly Hills, ASML, Google Cloud, Applied Intuition, Perplexity and Logical Intelligence executives said shortages could last two to five years.
Google Cloud said demand is surging, with backlog jumping from $250 billion to $460 billion, while energy limits are prompting exploration of orbital data centres and tighter hardware-software integration.
Panelists also highlighted real-world data shortages for physical AI, security guardrails for agents, sovereignty concerns over autonomous systems, and alternative models they say could outperform giant language systems in some tasks.
As nations demand 'AI sovereignty,' will the global AI supply chain fracture into competing geopolitical blocs?
Is AI’s insatiable energy demand creating a bubble that the world's power grids cannot sustain?
Are today's giant AI models a dead end, with smaller, rule-based systems being the only sustainable path forward?