Oracle CEO: GPU Oversupply Concerns Are Grossly Underestimated, AI Demand Far Exceeds Market Expectations
nashnova research
Oracle's CEO says market fears of a GPU oversupply rest on a fundamental misjudgment — real AI demand dwarfs current pricing, meaning the discount built around excess capacity may face a systematic re-rating.
What exactly did he say?
Oracle's CEO argues the risk of GPU oversupply is not worth worrying about over the next three to five years.
His core claim is one sentence: real AI demand is far larger than the market currently expects.
This means → He is not saying "supply is fine." He is saying demand has been severely underestimated — the imbalance sits on the demand side of the scale.
Why would AI demand keep growing?
AI adoption has already broken through in software development — that is his first anchor point.
He then points to mathematics, science, and adjacent fields, arguing they will spawn entirely new industries and drive sharp growth in computing demand.
In plain terms = What we see today is the first few pages of the book; many chapters remain unopened, and each one needs more GPUs.
What does this mean for the market?
The market's fear of GPU overcapacity is already priced in as a discount on related stocks.
If AI demand expands as broadly as he describes, that discount faces re-pricing pressure.
This means → The change would not be in GPU companies' fundamentals — it would be that the market drew the demand ceiling too low.
This reflects a deeper disagreement: the market is pricing "enough," while Oracle is saying "nowhere near enough."
市场有风险，内容仅供研究参考，不构成投资建议。