AI Data Center Power Shortage Puts Chip Supply Chains at Risk
A Morgan Stanley estimate cited in the article puts the US data-centre power shortfall at about 32GW by 2028, or roughly 34% of projected demand, even after planned mitigation measures. The constraint is increasingly electricity access and grid connections, rather than chip supply alone. Lawrence Berkeley National Laboratory data cited in the article show more than 2,060GW of generation and storage capacity awaiting grid connection at the end of 2025; connection processes can take five to seven years.
Morgan Stanley sees Nvidia and Broadcom as relatively better protected because customers can prioritise core accelerators and networking equipment, redirect deployments to locations with available power, and coordinate with infrastructure providers. Memory, optical components, power-management and analogue-chip suppliers may face greater order delays and inventory swings because their shipments depend on full data-centre systems coming online. Server makers sit between the two groups: large backlogs may not translate into revenue on schedule.
The article argues that power constraints are more likely to delay AI infrastructure spending than eliminate demand. For traders, that makes delivery timing, customer inventory and grid approvals key risks for quarterly earnings and valuations. Electricity infrastructure, including transformers, generation and cooling, may benefit from sustained investment, but project delays remain a risk across the supply chain.
Neutral
The article has no direct catalyst for cryptocurrency prices: it reports on US data-centre power constraints and the differing exposure of semiconductor and infrastructure suppliers. The crypto-market link is indirect. AI data centres and crypto mining both require substantial, reliable electricity, so prolonged grid queues or competition for power could raise costs and constrain new capacity for energy-intensive mining operations. That could affect miners’ margins, expansion plans and, in some regions, the economics of hosting or selling power to data centres.
In the short term, the report may encourage traders to monitor energy prices, grid policy and power availability when assessing publicly listed miners and infrastructure-linked assets. It could also contribute to broader risk sentiment if delays lead investors to reassess AI capital expenditure, although the article describes delays rather than a collapse in demand. In the longer term, investment in generation, transmission, storage and cooling could expand the pool of infrastructure available to both data centres and miners. The outcome depends on project delivery and regulation, not a clear change in crypto fundamentals.
Past episodes of rising energy costs have tended to affect mining profitability and the relative competitiveness of operators, but this article provides no evidence of a specific change in crypto prices, hashrate or network security. With no direct crypto asset or market data cited, a neutral classification is most appropriate; traders should treat the implications as a sector-level risk factor rather than a directional crypto signal.