'Nvidia is Boeing': Chamath Palihapitiya and Ken Griffin Back Viral Analogy Comparing AI GPUs to Airlines
Venture capitalist Chamath Palihapitiya and Citadel LLC Founder and CEO Ken Griffin supported a financial framework that equates the artificial intelligence (AI) GPU market to the commercial airline industry. The model compares GPU manufacturers to aircraft builders and AI compute capacity to airline inventory.
The AI Aviation Model
The analogy gained public traction following a publication by Alex Song, co-founder of Made Card and former asset-backed securities trader, who wrote, “A GPU neocloud has basically the same operating model as an airline.” Song asserted that hardware manufacturers control the market surplus, stating that “Nvidia [Nvidia Corp. (NASDAQ: NVDA)] is Boeing [Boeing Co. (NYSE: BA)] and Airbus rolled into one.”
According to Song’s framework, AI cloud providers lock in high fixed infrastructure costs while selling compute on demand, leaving profitability entirely dependent on hardware utilization.
Endorsements from Executives
On Wednesday, Palihapitiya shared Song’s post on the social media platform X, stating it is a “Super interesting comparison of GPU finance and the AI buildout to airlines.”
Palihapitiya noted that Griffin and Snowflake Inc. (NYSE:SNOW) CEO Sridhar Ramaswamy had made the identical comparison, adding, “Both are very, very smart so worth learning more.”
Griffin previously outlined the economic risks of this structure. According to reports of his Oct. 21 conference appearance, Griffin stated, “GPUs, like airlines, have a very low marginal cost to run,” warning that “the moment you have overcapacity, pricing goes into freefall.”
Perishability and Financing Risks
Ramaswamy also detailed the similarities in an Oct. 28 social media thread. Comparing AI compute to flight inventory, Ramaswamy stated that “airlines lease planes for years and sell seats one flight at a time.” He noted the perishability of both assets, writing, “an empty seat at takeoff is gone forever. so is yesterday’s unused GPU-hour.”
Song’s analysis further compares data centers to airframes and GPUs to aircraft engines. According to Song, a debt-financed GPU cluster requires “roughly 65% to 75% utilization” to cover fixed costs. Ramaswamy concluded his industry analysis by warning that due to this financial setup, “a capacity bust is more likely than a bank run.”
Here’s a list of the performance of a few semiconductor ETFs that track GPU companies for investors to consider.
| Stocks | YTD Performance | Six Months Performance | One Year Performance |
| iShares Semiconductor ETF (NASDAQ:SOXX) | 73.32% | 15.99% | 68.59% |
| VanEck Semiconductor ETF (NASDAQ:SMH) | 57.63% | 13.63% | 54.00% |
| Invesco PHLX Semiconductor ETF (NASDAQ:SOXQ) | 64.91% | 13.78% | 59.42% |
| SPDR S&P Semiconductor ETF (NYSE:XSD) | 52.19% | 5.21% | 38.43% |
| Direxion Daily Semiconductor Bull 3X ETF (NYSE:SOXL) | 155.41% | -9.00% | 117.40% |
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
Image via Shutterstock