Short-Seller Jim Chanos Scrutinizes Jensen Huang-Led Nvidia's AI Factory Cost Estimates: 'Well Above What Companies Are Telling Investors'
Rishabh Mishra 1 year ago
Analyst ColorEquitiesMarket SummaryNewsSector ETFsShort SellersBroad U.S. Equity ETFsShort IdeasFuturesMarketsTechTrading IdeasETFsGeneral
Short-seller Jim Chanos challenges NVIDIA's AI factory costs, warning estimates are "well above" what data center firms tell investors.
Just a day after Nvidia Corp. (NASDAQ:NVDA) and OpenAI announced a landmark $100 billion AI infrastructure deal, famed short-seller Jim Chanos is raising red flags over the foundational economics of the AI gold rush.
In a post on X, Chanos questioned NVDA CEO Jensen Huang's cost estimates for building a large-scale AI data center, suggesting a significant discrepancy with what other industry players are reporting.
Is Nvidia Inflating The AI Factory Cost?
Chanos highlighted Huang’s projection that a one-gigawatt (1GW) “AI factory” would cost between $20 to $30 billion before the cost of GPUs is even factored in.
He noted this figure is “well above what many AI data center companies are currently telling investors their costs will be,” pointing to a potential disconnect in the true price of building the infrastructure for artificial intelligence.
The critique centers on what is now being called “Jensen’s math,” which estimates a total 1GW facility cost of $60-$80 billion, with $40-$50 billion of that representing the “compute cost,” which is NVIDIA’s potential revenue.
Jensen's estimate that a 1GW AI factory will cost $20-30B BEFORE GPU costs, is well above what many AI data center companies currently are telling investors their costs will be. https://t.co/PcH5F6QMbS
— James Chanos (@RealJimChanos) September 22, 2025
See Also: Nvidia Pledges $100 Billion To Supercharge OpenAI: ‘This Is A Giant Project,’ Huang Says
Nvidia Pledges $100 Billion To Supercharge OpenAI
The scrutiny comes at a pivotal moment. The new NVIDIA-OpenAI partnership aims to deploy at least 10 gigawatts of AI systems, a project Huang himself called “giant.”
Based on Huang's own prior statements that “every gigawatt is about $40 billion, $50 billion to Nvidia,” the deal underscores the astronomical capital required for the current AI build-out.
What Is The Projected Data Center Capacity In The Coming Years?
Asset management giant Brookfield’s report projects that total AI data center capacity will surge from 7 GW in 2024 to 82 GW by 2034, a 28% compound annual growth rate. –
I'll take the "under". https://t.co/MzKZVsBGIc
— James Chanos (@RealJimChanos) September 14, 2025
Similarly, the McKinsey report shows total data center capacity demand, including non-AI workloads, growing from 82 GW in 2025 to 219 GW by 2030. Overall, McKinsey anticipates that capital spending on data center infrastructure, excluding the IT hardware itself, will surpass $1.7 trillion by 2030.
Whereas, when only considering the AI workloads, the capacity is expected to expand from 44GW to 156GW from 2025 to 2030.
According to the calculation based on "Jensen's math," the gigawatt demand translates into a staggering $6.2 trillion market opportunity for NVIDIA. This figure comes from multiplying 156GW capacity by 2030 with the $40 billion figure stated by Huang.
Even if that estimate is twice as high as reality, the post notes the opportunity would still stand at a formidable $3.1 trillion.
AI Data Center demand to grow from 44GW to 156GW from 2025 to 2030.
— The AI Investor (@The_AI_Investor) September 15, 2025
If every 1GW ~ $40B in market (Jensen's math) then Nvidia opportunity is going to be $6.2T, now if Jensen's math was 2x too high then it's still going to be $3.1T.
Either McKinsey & Company was too bullish on… https://t.co/i2c2Zu0pSn pic.twitter.com/8tLtK6ushr
Who Is Right When It Comes To AI Data Center Capacity?
Chanos’s skepticism introduces a critical question for investors: Who is right? If Huang’s non-GPU cost estimates are accurate, it could imply that data center operators and other infrastructure companies are underestimating future capital expenditures, potentially squeezing their margins.
This casts a shadow of financial reality over the industry’s explosive growth narrative, suggesting the true cost to power the AI revolution may be even higher than anticipated.
Price Action
Here is a list of AI-linked instruments that investors can consider;
| ETF Name | YTD Performance | One Year Performance |
| iShares US Technology ETF (NYSE:IYW) | 23.48% | 31.99% |
| Fidelity MSCI Information Technology Index ETF (NYSE:FTEC) | 21.03% | 30.62% |
| First Trust Dow Jones Internet Index Fund (NYSE:FDN) | 17.62% | 36.01% |
| iShares Expanded Tech Sector ETF (NYSE:IGM) | 24.85% | 34.99% |
| iShares Global Tech ETF (NYSE:IXN) | 22.26% | 26.71% |
| Defiance Quantum ETF (NASDAQ:QTUM) | 28.98% | 72.83% |
| Roundhill Magnificent Seven ETF (BATS:MAGS) | 21.09% | 39.95% |
The SPDR S&P 500 ETF Trust (NYSE:SPY) and Invesco QQQ Trust ETF (NASDAQ:QQQ), which track the S&P 500 index and Nasdaq 100 index, respectively, rose slightly in premarket on Tuesday. The SPY was up 0.015% at $666.94, while the QQQ rose 0.071% to $602.63, according to Benzinga Pro data.
Read Next:
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga
Photo: Shutterstock