The controversy surrounding Jensen Huang, CEO of NVIDIA, is affecting the company's operations in China and its financing for AI data centers. Analysts suggest that the investigation into Huang's extramarital affair with Alice Huang could further restrict NVIDIA's shipments to China. At the same time, borrowing costs for NeoCloud, which has relied on NVIDIA for funding and chip supplies, may also rise.
As of November 9, approximately 600,000 units of NVIDIA's H200 AI semiconductors are stuck in Hong Kong and unable to reach mainland China. Although the U.S. Department of Commerce relaxed export review criteria for the H200 in January, actual sales have not occurred. Commerce Secretary Gina Raimondo stated that the Chinese government is blocking imports to bolster its domestic semiconductor industry.
The remaining inventory is also expected to face shipping difficulties. U.S. lawmakers have already targeted NVIDIA's chip exports to China.
Senators Elizabeth Warren and Jim Banks requested an investigation in March into whether Huang's statement that there was "no evidence of AI chips being diverted" influenced export licensing decisions. This request came shortly after officials from Supermicro were indicted for allegedly smuggling billions of dollars' worth of AI servers to China.
With Alice Huang, a former executive at a company facing similar allegations of chip circumvention, being close to CEO Jensen Huang, Congress now has grounds to revisit existing export licenses. If the Chinese import restrictions are compounded by pressure from U.S. lawmakers to review export approvals, NVIDIA's ability to convert its inventory in Hong Kong into sales in China will be further diminished.
NVIDIA has already excluded China from its revenue forecasts. Colette Kress, NVIDIA's Chief Financial Officer, stated during the August earnings call that "due to geopolitical uncertainties, we did not include revenue from Chinese data centers in our future outlook." While resuming H200 shipments could have added previously unanticipated revenue, this situation has made that prospect more distant.
The burden on NVIDIA is likely to shift to the AI data center industry. NVIDIA serves as both a chip supplier and investor for NeoCloud's funding. The company participated in a $3 billion private placement led by CoreWeave alongside Jane Street.
According to overseas analysis firm Fork MBA, CoreWeave's credit default swap (CDS) costs are approximately 8.3 times that of NVIDIA. With the market already viewing NeoCloud's credit as significantly weaker than NVIDIA's, any instability at NVIDIA could lead to increased borrowing costs for NeoCloud.
NVIDIA also has a substantial backlog of orders in its supply chain. As of the end of July, NVIDIA's long-term supply and production capacity contracts totaled $279 billion. Relying on contract manufacturing without its own factories, NVIDIA entered these agreements anticipating increased demand. If demand forecasts decline, these contracts could become a financial burden for NVIDIA.
Funding timelines for AI companies are also concentrated in the same week. Anthropic is set to begin its IPO roadshow this week, aiming for a Nasdaq listing on November 16, just one day before NVIDIA's earnings report. OpenAI has raised $11 billion at a valuation of $1.25 trillion and plans to present its third-quarter results to investors this week. The uncertainty stemming from NVIDIA coincides directly with the funding timelines for AI companies.
The critical moment will come on November 17 when NVIDIA reports its third-quarter results for fiscal year 2027 (August to October). Market consensus estimates revenue at approximately $109 billion and adjusted earnings per share (EPS) at about $2.47. If the results fall short of expectations or if the fourth-quarter outlook is weak, concerns surrounding the CEO controversy and issues with China exports could signal a slowdown in performance. This could disrupt the funding that the AI data center industry has been anticipating from NVIDIA.
Domestic memory companies, including Samsung Electronics and SK Hynix, which supply high-bandwidth memory (HBM) to NVIDIA, are also affected by NVIDIA's performance and outlook, as it will influence their supply volumes and pricing negotiations for next year.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.