NVIDIA's initiative to partner with six major Wall Street firms in creating an AI infrastructure financing platform has triggered sharply divergent reactions across capital markets. While shares of alternative asset management giants surged, technology stocks continued to face downward pressure for a second consecutive day, highlighting deep market uncertainty about this financing model.

On Tuesday, the alternative asset managers involved in the partnership emerged as the biggest winners. KKR surged 6.88% to lead the pack, Apollo Global Management rose 6.26%, Brookfield Asset Management gained 4.77%, Blackstone climbed 3.89%, and BlackRock advanced 1.54%.

GPU cloud computing provider CoreWeave rose 2.42% during regular trading hours and surged more than 16% in after-hours trading. The market interpreted this financing arrangement as a direct positive for private credit and alternative investments, with the massive capital pipeline signaling long-term growth in management fee revenue.

Meanwhile, large-cap technology stocks weakened for a second straight day. Google fell 3.84%, recording its largest single-day drop in nearly six months. Amazon declined 2.09%, while NVIDIA itself barely moved, edging down just 0.02%. All three major U.S. stock indexes fell for the second consecutive day, with the Nasdaq dropping 0.6%.

The core market concern is whether NVIDIA's need to personally help customers find financing indicates that AI computing demand is so strong it requires financial innovation to be met, or whether certain demand itself needs financing conditions to be "created."

Huang's "Big Idea": AI Infrastructure as a New Asset Class

On Monday, NVIDIA CEO Jensen Huang appeared on CNBC alongside executives from Goldman Sachs, BlackRock, Blackstone, KKR, Apollo Global Management, and Brookfield Asset Management. They announced that the six institutions had signed memorandums of understanding to establish independent computing financing platforms, raising funds through third-party capital markets for AI infrastructure construction. The target is $500 billion, potentially higher.

Huang called this plan a "big idea," with its core logic being the redefinition of AI computing power as an investable asset. "These systems are not like our PCs, not like our phones," Huang told CNBC. "They are now revenue-generating assets, productive, long-lived, fungible, and flexible."

KKR's digital infrastructure head Waldemar Szlezak further explained the financial implications of this logic: "You can think of it as a revenue stream, then securitize it, or effectively split the risk and sell it to investors who want to participate at any level."

Goldman Sachs CEO David Solomon added: "You are starting to see asset-backed financing targeting this infrastructure construction. This is not surprising because these are real assets with real value."

Under the arrangement, the six financial institutions will make independent lending decisions. NVIDIA will connect customers with financing partners and can choose to assume up to 25% of the guarantee exposure for each loan. Huang later clarified on social media platform X that this support is "based on residual value, intended to supplement rather than replace independent underwriting."

The day after the news broke, shares of the participating alternative asset managers became the most prominent winners on the trading floor. The market logic is clear: the massive capital pipeline directly corresponds to the long-term growth outlook for management fee income.

At the same time, "new cloud" service providers also received a boost. NEBIUS rose 4.95%, RIOT gained 4.33%, Hut 8 climbed 3.64%, and IREN advanced 2.61%.

The Philadelphia Semiconductor Index closed up 0.87%, with 24 of its 30 components rising. Entegris rose 4.16%, KLA Corporation gained 4.01%, Teradyne climbed 3.96%, and ASML advanced 3.8%. Memory concept stocks mostly moved higher, with SK Hynix rising 4.7%, SanDisk gaining 2.68%, and Seagate Technology climbing 2.44%. Optical communication concept stocks generally strengthened, with CRDO rising 3.23% and Marvell Technology gaining 1.8%.

CoreWeave's after-hours surge was particularly noteworthy. As a GPU cloud computing provider, CoreWeave is the most direct beneficiary of this financing model. It represents the "new cloud" companies that have real computing demand but lack the balance sheet support of hyperscale cloud providers, exactly the target customer base NVIDIA's financing platform intends to serve.

Tech Giants Under Pressure, Market Divergence Sharpens

In stark contrast to the strength of the computing infrastructure chain, large-cap technology stocks weakened for a second consecutive day.

Google fell 3.84%, and since announcing its AI department restructuring last week, it has declined on four of the past five trading days, dragging down the communication services sector as the worst performer among the S&P 500's 11 sectors. Amazon dropped 2.09%, Apple fell 1.09%, Broadcom declined 1.5%, and Microsoft slipped 0.44%.

This divergence suggests deep market concerns. Asset managers surged on expectations of higher management fees, but tech giants and NVIDIA's own stock failed to benefit.

The underlying logic is that if AI computing demand were truly that strong, hyperscale cloud providers would be the most direct beneficiaries. However, NVIDIA's need to personally build financing channels for customers has instead raised concerns among some investors about whether potential customers' balance sheets are already under pressure.

Notably, NVIDIA's five-year credit default swap (CDS) has surged about 90% this year. Although it narrowed by 5 basis points to 72.11 basis points on Tuesday following Huang's clarification, it remains near historical highs, indicating that credit markets have not fully digested the "revolving financing" concerns.

The "Revolving Financing" Debate: GPU Residual Value is the Core Mystery

The core logic of NVIDIA's financing platform is not complicated: the bottleneck in the AI industry is shifting from chips and electricity to capital.

New customers outside the hyperscale cloud providers—AI labs, emerging cloud computing companies, sovereign AI projects—have genuine computing demand but lack the balance sheets of Google and Microsoft. A 1 GW AI data center costs about $50 billion, and OpenAI still does not have an investment-grade credit rating.

BlackRock CEO Larry Fink compared this model to the early stages of the mortgage-backed securities market: "This is a very early stage, just like when I started in the mortgage-backed securities market in the 1970s. I see this as the next future of financial engineering."

Apollo Global Management President Jim Zelter acknowledged that some risk exists: "There will be excesses, there will be corrections." But he added that the large number of participants helps diversify concentration risk.

However, the market's most pressing question remains unanswered: what is the long-term residual value of GPUs as collateral?

NVIDIA's own chip iteration cycle has accelerated from every two years to every year, with each generation's performance leap diminishing the market value of the previous generation. The foundation of the financing model—chip residual value—has not yet been tested through a full cycle.

Additionally, the announcement was merely a memorandum of understanding with no contractual obligations. Specific borrowers, interest rate levels, facility locations, and start dates have not been disclosed. The lack of detail has kept some investors on the sidelines.

Brookfield CEO Bruce Flatt offered a broader perspective: "Huang is leading the creation of these structures because there are trillions of dollars in the world." This may be the strongest support for this "big idea," but also its greatest unknown.