Overview

• Deutsche Bank’s George Saravelos argues the U.S. would be near recession without this year’s AI-driven capital spending and says the pace required to sustain GDP support is unlikely to continue.
• Bain & Company estimates the sector needs about $2 trillion in annual revenue by 2030 to fund projected compute demand and calculates an $800 billion gap even after efficiency savings.
• Bain highlights four scaling bottlenecks—power, construction capacity, GPUs, and ancillary gear such as switchgear and cooling—that could slow or raise the cost of datacenter buildouts.
• Nvidia unveiled a roughly $100 billion investment in OpenAI to develop at least 10 gigawatts of AI datacenters, underscoring how major vendors are still expanding infrastructure.
• Evidence of weak returns is mounting, with a recent study finding 95% of U.S. generative AI projects produced no measurable revenue and industry veterans warning many startups lack basic operational data practices.