Jim Chanos Questions Jensen Huang Over Why Nvidia Doesn't Rent Out Its Own AI Chips
Yahoo Finance ·
Prominent short seller Jim Chanos has publicly challenged Nvidia Corp. CEO Jensen Huang after the chief executive described the company's computing hardware as durable and highly rentable. Replying on social media, Chanos asked why Nvidia does not rent out the chips directly or simply continue raising prices. The exchange followed market data showing that hourly rental rates for three-year-old H100 training chips climbed 22% in a month to $3.28. Chanos has previously criticized data center operators like Oracle Corp. and CoreWeave Inc. for using unrealistic six-year depreciation schedules, warning that equipment can become obsolete in three to four years, creating massive financial risks. Hedge fund manager Michael Burry also estimated that cloud providers could understate depreciation by $176 billion from 2026 to 2028. Meanwhile, Nvidia recently partnered with financial institutions including BlackRock, Blackstone, and Apollo Global Management to secure over $500 billion for data center financing. At last check, NVDA stock closed 2.01% higher at $225.73 per share on Friday, showing robust gains across multiple timeframes.
AI 시장 분석
Short-seller Jim Chanos publicly questioned Nvidia CEO Jensen Huang on why they do not directly lease older AI chips, raising depreciation accounting controversies. The market is focusing on the retention of older hardware value as Nvidia's H100 training chip rental prices rose 22% in a month to $3.28 per hour. However, Michael Burry and Chanos warn that cloud companies' six-year depreciation periods are excessive, with actual chip lifespans being only three to four years, which could pose massive financial risks. Investors must closely examine the actual economic lifespan of assets and the sustainability of equipment rental profitability amidst the AI infrastructure investment boom.
상승 영향
- AI — Rental prices for three-year-old H100 chips rose 22% in a month to $3.28 per hour, showing that older AI hardware retains higher economic value and residual life than expected, proving sustained demand for AI computing infrastructure.
하락 영향
- Cloud — Major data center and cloud operators like Oracle and CoreWeave apply unrealistic six-year depreciation schedules, understating about $176 billion in costs between 2026 and 2028, and face massive financial risks from premature obsolescence within three to four years.
AI가 생성한 분석으로 투자 자문이 아닙니다.
DYAX Investor Sentiment
Bullish (Long) 52% · Bearish (Short) 48%
380 participants
Related News
- AMD Rises 3% as CFO Lifts AI Chip Market Forecast Toward $3 Trillion; Intel Ticks Up
- Here's Exactly What Drove BYD's 30% Spike in Profits -- Is It Still a Smart Buy?
- Memory Is the New ‘King’ of Chips, Taking Over 50% of Semiconductor Revenue, Susquehanna Says
- Pure-Play AI Cloud vs. Energy-Driven Infrastructure: Nebius Scale Meets IREN Limited’s Efficiency
- Forget Smartphones: Qualcomm Just Landed a Massive AI Deal With Amazon
- Should You Grab Nebius In Anticipation of Full Capacity Pricing?
