Morgan Stanley: AI Computing Power Gap Widens, Five Major Directions Emerge
TechFlow news, according to TechFlow Research, Morgan Stanley's Global Thematic Strategy Report on July 27 pointed out that the recent pullback in the AI infrastructure sector is driven by sentiment rather than a fundamental turnaround, the supply-demand gap in AI computing power will continue to widen, and this adjustment presents an allocation window.
The report outlines five major investment directions: AI Infrastructure Bottlenecks (power supply providers such as Talen Energy, Vistra, Bloom Energy, Hut 8, etc.), Computing Power Manufacturing Ecosystem (NVIDIA, Broadcom, Micron), China AI Solutions (ByteDance, Alibaba, Tencent accelerating capital expenditure, AI chip self-sufficiency rate expected to rise from 42% to 70%), Energy Security (South Korean power equipment manufacturers benefit from US transformer shortage, HD Hyundai Electric, LS Electric), Hyperscalers (Meta, Google, Microsoft, Amazon, AI capital expenditure ROI underestimated).
Google plans to expand computing power by 1000 times in 4 to 5 years, while NVIDIA's shipment CAGR during the same period is only 140%, the scissors gap is difficult to close in the short term. Enterprise Token monthly consumption is only $11, single project AI transformation saves $55, ROI exceeds 20 times, budget caps represent phased adjustments rather than a demand inflection point. US data center power gap reaches 38GW from 2026 to 2028, 14 states across the US are drafting control bills. Morgan Stanley suggests adding positions in batches when the theme retraces by 5% and 15%.
