[New York = E-Daily Seong Joowon Correspondent] Controversy over “circular financing” is reigniting amid a series of massive artificial intelligence (AI) investment and financial support deals recently finalized by NVIDIA. Warnings are growing that industry demand and valuations are being artificially inflated as a pattern repeats in which companies in which NVIDIA invests or provides funding ultimately end up purchasing NVIDIA chips.

At the heart of the controversy are a series of new deals totaling $750 billion (approximately 1,103 trillion won) that NVIDIA recently announced. According to Bloomberg on the 27th (local time), NVIDIA unveiled an AI initiative worth more than $500 billion on the 24th with the SK Group—including SK Telecom ( SK hynix(000660)), the largest shareholder of SK Telecom, and SK Innovation ( SKSQUARE(402340)). At the same time, discussions are underway to provide up to $250 billion in guarantees to enable OpenAI to lease a data center hub in Ohio, U.S. (valued at $500 billion and with a capacity of 10 gigawatts) owned by a SoftBank Group subsidiary, as well as to provide financial support for OpenAI’s purchase of its own chips worth $350 billion. However, Bloomberg reported that negotiations regarding OpenAI are in the early stages and could fall through or see their terms change.

“Investment Equals Revenue”… Growing Warnings About Circular Financing

Gary Tan, a portfolio manager at Allspring Global Investments, noted,

“

While NVIDIA’s investments and partnerships reinforce confidence in long-term AI development, investors are concerned

about circular financing,”

adding, “Capital is increasingly being used to fund future AI customers and infrastructure deployments.” Billy Leung, an investment strategist at Global X Management, also noted, “NVIDIA’s additional guarantee of OpenAI’s data center debt exacerbates vendor financing, which is already under scrutiny,” adding, “This is both a signal of demand and evidence of the funding pressures involved in the AI deployment process.”
Transactions with similar structures are not unique to NVIDIA. Google also guaranteed rent payments for five of Anthropic’s data centers, effectively helping Anthropic secure a $35 billion loan. Warnings are emerging that as AI companies become increasingly intertwined in this way, the entire industry is being exposed to systemic risk. In particular, rising industry debt levels are cited as a key concern, as many AI companies are increasing their borrowing to fund data center and chip projects. NVIDIA has already announced similar transactions totaling more than $540 billion this year alone, excluding new discussions with OpenAI.
On the other hand, Jensen Huang, CEO of NVIDIA, has consistently dismissed such claims of a “circular financing” cycle. Referring to the investment in CoreWeave last January, he countered, “This represents a small fraction of the total funding these companies will ultimately need to raise,” adding, “The notion that this is a circular cycle is preposterous.”
However, the market’s reaction was sobering. NVIDIA’s credit default swap (CDS) premium—a derivative that acts as insurance against the risk of bond default—soared by its largest margin on record that day. According to ICE Data Services, the cost of insuring against default on NVIDIA’s 5-year bonds surged to 0.82 percentage points annually, up 0.14 percentage points at one point during the trading session. This marks the largest intraday increase since November of last year, when trading in these swaps began in earnest. This indicates that concerns over an increase in NVIDIA’s debt burden were immediately reflected in bond prices right after news broke of negotiations for a $750 billion AI infrastructure deal.

Tangible Benefits for Korean Semiconductor and Platform Companies

Regardless of this controversy, the deal is directly benefiting Korean companies. SK hynix and SamsungElectronics(005930)currently split the market for high-bandwidth memory (HBM—high-performance memory semiconductors designed to quickly process data required for AI computations), and NVIDIA plans to support SK hynix’s next-generation HBM design to ensure a stable supply chain. The two companies are also jointly building an AI data center on the Korean Peninsula with a capacity of over 2 gigawatts (enough to power 1.5 million households), and the first “AI Factory” being constructed by SKTelecom(017670)is set to open next year. CEO Hwang explained, “Together with SK Group, we will be undertaking a business venture worth $500 billion.”
On the same day, NVIDIA announced that it would invest $1 billion in Naver to support the expansion of an AI data center currently under construction in South Korea (jointly developed with the U.S. private equity firm Brookfield). With this funding, Naver will be able to more than triple the size of the facility. On that day, Naver’s (NAVER(035420)) stock price surged by more than 8%.
Meanwhile, SoftBank Group Chairman Masayoshi Son continues to pursue his ambition to position the company at the center of the AI industry through OpenAI. As of October this year, SoftBank had committed approximately $65 billion in investments to OpenAI alone and secured a $40 billion bridge loan—one of the largest in the Asia-Pacific region—to fund this. However, as bets on companies with limited control grow, so does investor anxiety.
Amid this situation, the key question is whether concerns about a vicious cycle of financing will materialize into actual risks. Critics argue that if AI demand fails to meet current high expectations, the intertwined investment and financing structures could become a trigger that amplifies losses. The next key variables are whether the $250 billion guarantee and $350 billion purchase financing negotiations between NVIDIA and OpenAI will lead to actual contracts, and what kind of ripple effects the spread of such vendor financing will have across the industry.