Missed the webinar? Here are Investing.com’s top 10 stock picks for 2026
Investing.com -- In a note to clients on Monday, Morgan Stanley analyst Stephen Byrd warned that the rapid expansion of artificial intelligence is deepening a structural gap between the compute required to train advanced models and the power needed to run the data centers behind them.
The analyst stated that it is “again revising upward our data center power demand forecast, driven by higher chip demand,” citing a “non-linear rate of AI improvement” that is pushing energy requirements sharply higher.
The bank now projects a “47 GW cumulative shortfall in 2025-28 vs. 44 GW prior,” following updated semiconductor sales estimates tied to stronger volumes at Nvidia.
Morgan Stanley believes the underlying theme is “the proliferation of intelligence,” with AI adoption expanding faster than infrastructure can keep pace.
By 2026, the bank expects investors to shift focus to what it calls “intelligence bottlenecks,” obstacles such as power constraints, political approvals, labour shortages and shortages of specialised data center equipment.
Morgan Stanley remarked that these constraints will define the next phase of AI investing and shape which companies outperform.
Among potential winners, the bank highlights operators best positioned to “de-bottleneck data center growth,” especially those offering rapid “time to power” solutions.
Morgan Stanley pointed to natural gas turbines, siting at operational nuclear plants and fuel cells from Bloom Energy, which it said could deliver “5-8 GW through this period” and potentially more with rapid capacity expansion.
But the “fastest path” to new power, the bank argues, is the conversion of Bitcoin mining sites, which already control “almost 20 GW of large… sites that have a firm grid interconnection agreement.”
IREN is seen as a flagship example of the “new neocloud” model. “Under the ‘new neocloud’ model, most notably exhibited by IREN, the Bitcoin miner purchases GPUs/TPUs, builds the entire data center and leases the facility to hyperscalers and other customers - potentially under leases with relatively short durations (such as the 5-year lease signed by IREN with Microsoft),” wrote the bank.
Morgan Stanley said these locations offer “the lowest execution risk” for AI players and are likely to become increasingly valuable as the power gap widens.