The departure of Nvidia CEO transition and Chief Executive Jensen Huang has raised questions about whether the aggressive artificial intelligence investment strategy he championed will undergo changes. Observers note that the so-called ‘circular financing’ structureβin which Nvidia invests in AI companies that in turn purchase Nvidia’s graphics processing units (GPUs)βmay face a brake.
According to Nvidia, Colette Kress assumed the role of interim CEO following the resignation of the former CEO. Having served as Chief Financial Officer (CFO) since 2013, she is a finance expert with experience at Microsoft and Cisco. She is evaluated as a figure who managed the company’s financial strategies and investment risks while driving Nvidia’s rapid growth alongside the former CEO.
Particularly under Kress’s leadership, changes are expected in Nvidia’s aggressive circular financing structure. Nvidia has expanded its ecosystem by directly taking equity stakes in AI companies, guaranteeing cloud purchases, and providing credit enhancements for data center businesses. This structure interlocks investments and sales as funded companies purchase massive amounts of Nvidia GPUs. Concerns have continuously been raised that this could lead to an overestimation of AI demand.
The scale of investment is also substantial. As of the end of July, Nvidia’s equity investment valuation reached $99 billion (approx. 130 trillion KRW), with future equity investment commitments standing at $25 billion (approx. 33 trillion KRW). AI cloud capacity purchase commitments amount to $360 billion (approx. 48 trillion KRW). For the data center business to be utilized by OpenAI, the company decided to provide up to $105 billion (approx. 140 trillion KRW) in credit enhancements. It also signed a $6.3 billion (approx. 8 trillion KRW) contract with AI cloud provider CoreWeave to purchase cloud capacity that has not been sold to customers.
Kress is known to have maintained a cautious stance on such large-scale investments and guarantees. U.S. IT media outlet The Information reported that Kress opposed the $250 billion OpenAI credit enhancement initiative pushed by the former CEO. The actual contracted guarantee limit is currently $105 billion. Analysis suggests that if Kress applies stricter criteria to new investments and guarantees, market concerns surrounding circular financing could ease. On the other hand, reduced funding could also lead to a slowdown in AI infrastructure investments.
The appointment of the next CEO is also a variable. While the Nvidia board of directors has begun searching for a successor, foreign media have mentioned Ian Buck, Vice President of Nvidia who led CUDA development, and former Nvidia executive RenΓ© Haas, CEO of Arm, as major candidates. Buck is evaluated for his strengths in technological competitiveness succession, while Haas is noted for semiconductor expertise and large-scale corporate management experience. Former Apple CEO Tim Cook and AMD CEO Lisa Sue are also mentioned in the candidate pool, though their actual recruitment possibility remains opaque. The option of having a technical expert serve as CEO while Kress oversees financial strategies is also being discussed.
Following the departure of the former CEO, the ‘semiconductor cold war’ between the U.S. and China is showing signs of further intensification. The New York Times (NYT) reported that U.S. President Donald Trump tried to dissuade the former CEO from resigning. President Trump reportedly called him to urge him to fight to keep his position. President Trump is said to have been concerned about the negative impact that the departure of the former CEO would have on the U.S. AI semiconductor industry.
Taking advantage of this vacuum, China is urging South Korean semiconductor companies to target the Chinese market. The Chinese state-run Global Times argued in a commentary that the slowdown in U.S. data center construction could negatively impact South Korean semiconductor exports. It further asserted that to maintain and expand their foothold in the Chinese market, companies must supply cutting-edge products and find ways to bypass U.S. export controls on High Bandwidth Memory (HBM).
