Nvidia faces U.S. export restrictions on AI chips to certain markets despite strong global demand for its processors.
Investment firm Morgan Stanley has concluded that electricity demand will remain resilient regardless of potential recession stemming from Trump's tariff policies, primarily because data centers require substantial power and demand will not decline despite economic downturns or efficiency gains from companies like Ant Group and DeepSeek. The firm stated: "We believe power demand trends are more durable than in prior cycles, in part due to the inelasticity of data center demand. Industrial demand could decline in the near term, but reshoring of manufacturing is a long-term tailwind." A Bloomberg report forecasts that US power demand from data centers could surge 20-40% in 2025, with strong double-digit growth likely to persist through 2026-30. Morgan Stanley further projects electricity consumption from artificial intelligence growing tenfold by 2028, though the firm acknowledges that rapid policy changes may have deep implications for large capital investments.
Despite these positive long-term outlooks, Morgan Stanley cautioned that near-term risks exist. The firm noted: "With this in mind, we do not want to minimize the risk of a near-term 'shock' in demand. This could translate into slowing order growth for some companies." However, historical data supports the resilience of energy stocks, as demand has fallen by just 0.2% on average during economic downturns since 1960. Utilities are expected to fare well in a recession given their "defensive nature."
The analysis remains optimistic about strong infrastructure spending from major tech companies. Morgan Stanley is bullish on substantial capital investments from hyperscalers such as Meta, Amazon, and Alphabet, as these companies seek to maintain leadership positions in artificial intelligence and support large product pipelines that require GPUs.
NVIDIA Corporation (NASDAQ:NVDA), which specializes in AI-driven solutions for data centers, self-driving cars, robotics, and cloud services, ranks fourth on the list of top AI stocks in the spotlight. On April 17, Argus lowered the firm's price target to $150 from $175 while maintaining a "Buy" rating. According to Argus, new US licensing requirements for AI chip exports, including NVIDIA's H20 models, could impact quarterly earnings by as much as $5.5 billion. While the firm acknowledges NVDA's investment potential, it notes that some AI stocks have demonstrated greater promise for higher returns within shorter timeframes, with at least one AI stock rising since the beginning of 2025 while popular AI stocks declined around 25%.