Jensen Huang: NVIDIA Chips in Short Supply, Allocation Far Below Customer Demand
nashnova research
Jensen Huang publicly admitted Nvidia's chip allocations fall far short of customer demand, saying he is 'constantly different disappointing customers' — the supply bottleneck has shifted from chip fabrication to land, power, and physical infrastructure, with no near-term relief in sight.
How large is the demand?
Huang said new orders pour in almost every week, every day, at enormous scale.
He cited an Australian project — 2 gigawatts, roughly $80 billion — to illustrate the size of a single customer's needs.
This means → Nvidia's challenge is not whether to grow, but that growth cannot keep pace with the speed demand is expanding.
Where is the supply stuck?
Actual allocations fall far short of demand. The bottleneck is not chip design — it is land, power, and data-center infrastructure.
In plain terms = Nvidia can build the chips, but the data centers needed to run them have nowhere to be built and no power to draw on — so capacity simply cannot scale.
This reflects a shift in the AI compute race: the binding constraint is moving from "can the technology do it" to "are there enough physical resources."
Why is Huang going public with the numbers?
He said disclosing supply conditions openly is meant to prevent customers from suspecting they are being short-changed relative to rivals.
This means → Competition for compute resources has grown so intense that the vendor itself must step in to manage expectations.
The key variable ahead: whether infrastructure buildout can accelerate will directly determine Nvidia's capacity-release pace and the length of the customer queue.
市场有风险，内容仅供研究参考，不构成投资建议。