Nvidia SK AI Buildout Meets Power-Bill Pushback

AI infrastructure is moving from chip roadmaps into power contracts, memory supply, and voter bills.

🗄️ Nvidia and SK Put a $500 Billion Marker on AI Factories

Decoded: Reuters reported that Nvidia and South Korea's SK Group unveiled a more than $500 billion AI initiative covering large-scale AI data centers and next-generation memory. The plan includes a long-term SK Hynix memory partnership, HBM4 development for AI training and agents, and a 2-gigawatt SK Telecom AI data center powered by Nvidia Vera Rubin chips, with the first facility due in 2027. Nvidia's own Vera Rubin materials say NVL72 systems combine 72 Rubin GPUs and 36 Vera CPUs, and target one-tenth the cost per million tokens versus GB200 NVL72 for deep-reasoning inference.

Why it matters: This ties Nvidia's next growth leg to HBM availability, sovereign AI capacity, and power-scale deployments rather than chip demand alone. Investors should watch whether memory supply and grid access become the real gating items for AI infrastructure revenue.

🏛️ AI Power Politics Hit the Utility Bill

Decoded: Reuters reported that President Donald Trump announced a non-binding pledge by U.S. power producers and data center companies to fund or build the energy infrastructure needed for AI demand while limiting consumer cost increases. The same report cited Energy Information Administration forecasts for average residential electricity prices to rise 5.1% in 2026 and 2.4% in 2027 before inflation, and said data centers accounted for $6.3 billion, nearly 40%, of PJM's latest capacity-market charges.

Why it matters: The AI trade is broadening from semiconductors into regulated utilities, transmission, and local politics. The pledge may help data center developers with approvals, but voluntary commitments leave utilities and ratepayers exposed if demand arrives faster than new generation and grid upgrades.

Stay decoded. See you tomorrow.

— The Get AI Decoded Team