Nvidia, Broadcom Shielded From US Power Crunch; Memory Makers Exposed: Morgan Stanley

Morgan Stanley expects power shortages to affect parts of the AI chip ecosystem unevenly, with memory, optical, power-management and analog suppliers facing greater inventory risks if data-centre deployments are delayed or customer orders are pushed back.

Advertisement
Read Time: 2 mins
The US data-centre industry could face a 32 GW net power shortfall through 2028.
Unsplash

Nvidia and Broadcom are unlikely to see their 2027 forecasts disrupted by growing power constraints at US data centres, even as electricity shortages threaten to slow parts of the broader artificial intelligence chip supply chain, Morgan Stanley said.

The brokerage estimates that US data-centre developers face a 34% net power shortfall through 2028, equivalent to around 32 gigawatts, even after accounting for alternatives such as behind-the-meter generation and fuel cells.

Advertisement

The rapid expansion of generative AI has driven billions of dollars into data-centre construction, but the sector's rising electricity requirements are increasingly colliding with limitations in US power availability.

Morgan Stanley and Goldman Sachs have both highlighted challenges facing the data-centre buildout, although their assessments differ on the likely near-term impact of political resistance.

Advertisement

Also Read | Jensen Huang's Net Worth Crosses $200 Billion; Becomes World's 7th-Richest Person

For Nvidia and Broadcom, Morgan Stanley sees relatively limited exposure to the bottleneck. The brokerage pointed to the companies' visibility into chip deployment, geographic expansion plans and coordination among data-centre operators, semiconductor suppliers and power providers.

However, a slower pace of data-centre deployment could create problems further down the chip supply chain. If customers cannot install computing capacity as planned, they could delay deliveries or cancel orders, potentially leaving suppliers with excess inventory.

Advertisement

Memory chips, optical components, power-management products and analog semiconductors could be among the most vulnerable segments, according to Morgan Stanley.

The assessment highlights a growing distinction within the AI hardware market: leading chip suppliers with greater visibility over large deployments may be better positioned to navigate infrastructure constraints, while secondary component makers could bear the impact if AI infrastructure projects are pushed back.

Morgan Stanley has previously warned that labour, electricity availability and political hurdles could constrain the US data-centre expansion.

Also Read | Man Charged With Illegally Shipping Nvidia Chips To China

Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.


Loading...