AI Infrastructure Stocks Outpace Nvidia as Power Becomes the Bottleneck - TechStock²

As artificial intelligence (AI) continues to expand, power infrastructure has emerged as a key bottleneck constraining growth in the sector. Recently, stocks related to AI infrastructure—particularly ... Read original →

The AI infrastructure race has shifted decisively from chips to power. With NVIDIA’s Blackwell systems exceeding 100kW per rack, legacy power and cooling architectures collapse, forcing data centers to rebuild their entire energy stack—from grid interconnection to liquid cooling. GE Vernova and Vertiv now dominate as providers of modular substations and high-voltage DC distribution, becoming de facto standards for new AI clusters. FERC’s 60-day ultimatum exposes the U.S. grid’s regulatory lag; regional operators must overhaul outdated interconnection rules or face bottlenecks that will cull smaller players and favor engineering giants with cross-state capabilities. Over the next 18 months, capex will pivot from GPU procurement to power redundancy and PUE compliance. Memory vendors like Micron risk marginalization unless they integrate advanced power-aware IP into AI server designs. Crucially, valuation metrics are migrating from 'compute density' to 'watts per useful operation,' redefining co-design between semiconductors and energy infrastructure.