또 증시 덮친 ‘AI 수익성’ 논란…GPU 감가상각비 급증 ‘위험 요소’

A growing concern in recent stock markets is the investment performance of artificial intelligence (AI) companies, particularly those in the cloud computing sector. While revenue targets have been missed, the speed of earning profits remains sluggish. This has led to a significant increase in financial costs due to rising interest rates, and the rapid depreciation of investments in graphic processing units (GPU) used in AI development.

On Wall Street, companies like CoreWeave, Nebula, IREN, and Oracle saw their stocks plummet by 8-15% compared to the previous day.

These companies have invested billions in AI technology through a method known as "rotating finance," which involves leasing high-performance GPUs to cloud service providers and AI development firms. According to market research firm Gartner, AI infrastructure investments are expected to reach approximately $750 billion between 2026 and 2030.

To achieve an appropriate return on investment (ROI) of 12%, $850 billion in annual sales will be required by 2030. However, current AI-related cloud computing revenues are estimated to reach a maximum of $360 billion.

Future Asset Securities Vice President, Kim Myung-soo, warns that if the current investment trend continues, significant asset depreciation risks could emerge in the next few years, potentially leading to more than 80% of AI infrastructure investments becoming independent businesses. In the U.S. bond market, CoreWeave's bond rates have surged to 15%, causing the company's financial status to deteriorate.

Financial analysts have also criticized CoreWeave's asset value being destroyed due to a margin structure where the company is valued at a loss. D.A. Davidson analyst Gill Rulli notes that despite borrowing funds at a 9% interest rate, CoreWeave only achieves a 1% return on investment, likening it to "borrowing money at 9% to buy U.S. Treasuries yielding 5%."

He warns that as the business expands, it could become increasingly detrimental to shareholder value. Technology expert Michael Burry, referenced in the film "Big Short," points out that the economic lifespan of GPUs used in AI learning, processing, and inference is only 2-3 years, while cloud companies like Google, Meta, and Microsoft artificially extend their asset value through bookkeeping to create a "profit bubble."