[Edaily Marketin Reporter LEE GEON-EOM ] An analysis suggests that the credit risk profile of the memory semiconductor and power equipment industries is fundamentally shifting as global artificial intelligence (AI) infrastructure investment relies heavily on external funding sources such as private credit. The analysis indicates that if funding dries up due to interest rate hikes or a capital market crunch, infrastructure orders could plummet sharply—regardless of demand for AI services.

Kim Jeong-hoon, Senior Analyst at Korea Credit Rating (KCR), made these remarks on the 10th at the seminar titled “AI, High Interest Rates, Real Estate: Reviewing the Changing Market Environment and Credit Risks,” held at the Korea Exchange Conference Hall in Yeouido, Seoul. Senior Analyst Kim analyzed that as AI infrastructure investment has recently become increasingly linked to financial markets, the memory industry’s business conditions may unfold in a different pattern than before.

In particular, he noted that memory demand is heavily influenced by capital market conditions as structured finance—such as asset-backed securities (ABS) backed by data center rental income and loans based on graphics processing unit (GPU) computing fees—is being utilized across the board. In fact, it is projected that approximately half of AI capital expenditures (CAPEX) from 2025 to 2028 will be financed by external funds.

Senior Analyst Kim explained, “In the past, large Big Tech companies with strong cash generation capabilities were the main customers, so the creditworthiness of client companies was not a focus of analysis.” He added, “Recently, however, speculative-grade companies such as CoreWeave have entered the AI ecosystem, bringing the funding environment to the forefront as a key variable.”

He further pointed out, “When capital markets tighten, data center investment slows down,” adding, “This can directly lead to a decline in orders for GPUs and memory.”

He also emphasized that long-term supply contracts do not fully eliminate earnings volatility in the memory industry. He explained that if customers’ funding sources dry up due to a deteriorating external financing environment, it is difficult to rule out the possibility that the terms of long-term contracts will be renegotiated. However, even assuming an extreme scenario of slowing demand, SK hynix(000660)is expected to maintain net cash of around 100 trillion won based on its accumulated financial reserves, thereby demonstrating its resilience.

While the power equipment industry is not immune to concerns about an AI-driven slowdown in demand, the short-term impact is expected to be limited compared to that on the memory semiconductor sector.

Choi Seon-young, a senior analyst at Hanshin Rating, analyzed, “For power equipment, the impact of AI investment is reflected indirectly through demand for power infrastructure, so the response is relatively slow and replacement cycles are long.”

Another positive factor is that data centers account for only about 5–6% of global power demand, and long-lead-time orders—such as the replacement of aging power grids and the integration of renewable energy—make up a significant portion of the order backlog.

In fact, of the 16.5 trillion won in new orders secured in 2025 by three major domestic power equipment companies— HD HYUNDAI ELECTRIC(267260), HYOSUNG HEAVY INDUSTRIES(298040), and #LS Electric—the volume of direct shipments to AI data centers is estimated to be approximately 1 trillion won. However, experts point out that the unique financial risk factors faced by each company require continuous monitoring.

Chief Analyst Choi stated, “For HD HYUNDAI ELECTRIC, we must examine its ability to meet funding needs arising from large-scale domestic and international capacity expansions and increased dividends, while for HYOSUNG HEAVY INDUSTRIES, we must monitor whether contingent liabilities in the construction sector materialize.” He added, “For LS ELECTRIC, high exposure to U.S. tariff policies and the company’s ability to control working capital burdens will be key factors in future credit assessments.”