AI Trends

Jensen Huang Reportedly Orchestrated the Shortage Himself: Pouring Money into AI Startups to Absorb Excess GPUs

Sina TechUpdated Nov 18 11:33

AI summary

NVIDIA is reportedly making large-scale investments in AI startups and small and mid-sized cloud providers, letting them use that money to buy back its own GPUs, in order to absorb chips it over-ordered from TSMC and prop up GPU prices. Amid internal conflict, CFO Colette Kress pushed to scale back up to $250 billion in compute guarantees to companies such as OpenAI, ultimately cutting it to about $125 billion. Analyst Ed Zitron estimates that at least $200 billion worth of NVIDIA AI GPUs sit idle in warehouses due to power shortages or a lack of end demand.

Body · AI translation

Add crowdsourcing promotion copy Add crowdsourcing promotion copy end Add App-exclusive traffic diversion

Fast Technology News, November 14: According to reports, after Jensen Huang stepped down as NVIDIA CEO, an internal investigation revealed another side to the GPU and memory shortage. NVIDIA had made large-scale investments in AI startups, driving these companies to buy its own GPUs and build data centers, thereby absorbing the chips it had over-ordered from TSMC.

The trigger for this operation was an "avalanche-style" inventory crisis. Huang privately admitted that as hyperscale cloud providers shifted to self-developed chips, the large number of chips NVIDIA had obtained by paying huge advance orders to TSMC was "coming at him like an avalanche." If they flowed directly into the market, GPU prices would face a sharp plunge.

To prop up prices, Huang chose to use NVIDIA's balance sheet to "transfuse blood" into AI startups and small and medium-sized cloud providers, letting them use this money to buy NVIDIA GPUs in return. In essence, this used artificially created "demand" to absorb excess supply, rather than letting the market clear naturally.

This aggressive "circular financing" strategy sparked fierce internal conflict. In August 2026, CFO Colette Kress pushed to reduce up to $250 billion in compute guarantees to companies such as OpenAI, and Huang was reportedly furious, calling her a "deer caught in the headlights." Kress ultimately capped the guarantees at about $125 billion, but the direction was not fundamentally reversed.

Analyst Ed Zitron previously estimated that at least $200 billion worth of NVIDIA AI GPUs were sitting idle in warehouses due to a lack of power or end demand. For PC gamers who have endured high GPU and memory prices for years, this report means that the "severe shortage" they perceived may have been artificially amplified.

[This article ends] If you need to reprint, please be sure to indicate the source: Fast Technology

Editor in charge: Black Tea article_adlist[

Report article content ]article_adlist Statement: Sina.com exclusive manuscript, unauthorized reproduction prohibited. tech_web_article_v2015_block_bottom Non-targeted 300*250 button 17/09 wenjing begin

Non-targeted 300*250 button end

Source

Sina Tech