- Dell COO Jeff Clarke forecasts that agentic AI workloads will keep memory and hard disk drive supply tight for more than five years, making the current infrastructure cycle structurally longer than previous hardware booms.
- Dell projects an 87-fold increase in inference token generation and expects AI to be 75% of data center demand by 2030, requiring an additional 200 gigawatts of power capacity.
- Clarke named DRAM the most significant bottleneck alongside NAND, disk drives and other semiconductors, with demand ahead of supply even for traditional servers and the broader AI supply chain running at full capacity.
- Dell shares rallied over 3% after Morgan Stanley raised the probability of its $756 bull case within 12 months following a meeting with Clarke, while keeping a Neutral rating and $511 price target.
- Dell carries a $95 billion backlog for AI servers and a raised full-year revenue outlook of approximately $192 billion, with investors treating recent weakness tied to insider sales as a buying opportunity.
Quotes
“agentic AI could lead to component shortages, specifically memory and hard disk drives, persisting for over five years”
“demand remains ahead of supply even for traditional servers”
“AI agents represent a structural expansion of Dell's total addressable market, not merely a temporary hardware upgrade cycle”