자본시장 · 리포트

Wccftech는 전 엔비디아 CEO 젠슨 황이 GPU 가격을 낮추는 대신 과잉 실리콘을 흡수하기 위해 AI 스타트업에 자금을 지원했다고 보도했으며, 이는 공급 부족이라는 서사에 반하는 주장이라고 전했다.

If accurate, Nvidia-linked startup financing to soak up surplus chips would mean vendor financing, not list-price cuts, is propping up demand, a key question for GPU cycle durability. Wait, I need to correct my output: the above is the untranslated source text, which I should not have returned. Here is the Korean translation: 사실이라면, 잉여 칩을 흡수하기 위한 Nvidia 연계 스타트업 금융은 정가 인하가 아닌 벤더 파이낸싱이 수요를 떠받치고 있다는 뜻이며, 이는 GPU 사이클의 지속력에 대한 핵심 쟁점이다.

업계 전문지Slicast · 2026년 11월 13일 19:23 UTC · 미국 · 출처: Wccftech

중요도 50

Now that NVIDIA's Jensen Huang is no longer the company's CEO, we have some insight into the forces behind the sharp rise in semiconductor and memory prices. According to the latest findings, the industry was led to believe there was an organic supply shortage. In reality, Huang was reportedly using NVIDIA's own balance sheet to finance AI startups and datacenter buildouts so they could absorb the company's over-ordered semiconductor shipments from TSMC.

NVIDIA faced an "avalanche" of excess TSMC silicon and memory, and the only way to curb the oversupply was to fund AI startups.

The revelations stem from a tense August 2026 confrontation between Huang and NVIDIA's longtime Chief Financial Officer, Colette Kress, who is now interim CEO. Because the former CEO was escalating the GPU maker's financial exposure, Kress pushed to rein in the company's "circular financing" commitments to smaller cloud providers and major AI firms such as OpenAI and Anthropic, according to The Wall Street Journal.

Initial discussions had floated underwriting roughly 8 gigawatts of OpenAI compute valued at up to $250 billion. Kress successfully scaled NVIDIA's signed guarantee cap down to roughly 4.25 gigawatts, or up to $125 billion. Huang reportedly belittled Kress for questioning the deals, calling her "a deer in the headlights."

To justify NVIDIA's need to financially underwrite OpenAI and smaller cloud operators so they would keep buying its GPUs, Huang privately acknowledged that major cloud giants were pivoting away from NVIDIA in favor of in-house custom chips. With hyperscalers actively reducing their reliance on "off-the-shelf parts," Huang said the massive volume of prepaid TSMC wafers and memory NVIDIA had committed to buying was "coming at him like an avalanche."

By funding datacenter expansion for AI startups, NVIDIA effectively created artificial demand to clear its own surplus silicon, rather than letting the supply flood the market and force GPU prices down. These internal pressures align with earlier assessments from industry analyst Ed Zitron, who estimated that at least $200 billion worth of NVIDIA AI GPUs are sitting idle in warehouses. The chips were purchased by cloud giants that lacked either the electrical grid capacity or immediate end-user demand to power them on. Amazon faced a similar dilemma.

Knowing the avalanche would hit NVIDIA sooner rather than later, the company reportedly contacted TSMC to wind down its upcoming chip orders. For PC builders and hardware enthusiasts who have endured years of steep GPU and RAM prices under the belief that components were severely scarce, these disclosures undermine the shortage narrative and expose the aggressive financial maneuvering used to keep hardware prices elevated.

Huang cannot be faulted for making these moves, since he was acting in the company's best interests. Still, for gamers and hardware owners who have been unable to find attractive GPU or memory deals amid the perceived shortage, it feels like a stab in the back.

원문 보기