From Selling Chips to Providing AI Infrastructure Guarantees! Nvidia (NVDA.US) Plans to Back New Cloud Player Nscale to Support Anthropic's Computing Power Expansion

According to Zhitong Finance APP, "AI chip superpower" Nvidia (NVDA.US) is actively negotiating to provide major credit support for the new cloud force Nscale's large-scale AI computing infrastructure project. According to a source familiar with the matter, Nvidia is discussing providing a credit guarantee for a large-scale AI data center-type computing infrastructure of Nscale. Anthropic, the developer of the globally popular Claude series of AI applications and one of the leading companies in global AI applications, plans to use this facility.

The source, who requested anonymity due to the confidential nature of the discussions, said that this credit support would help secure power and construction costs for an AI data center park in West Virginia, which will be equipped with Nvidia-led cutting-edge AI GPU clusters. As the dominant supplier in the field of AI processors, Nvidia has already invested in Nscale and other AI computing service providers.

UK-based Nscale is accelerating the rollout of a series of large data centers, including this flagship park in West Virginia. The company is also moving forward with an initial public offering (IPO), though some investors have raised concerns about its $30 billion IPO target valuation.

A representative from Nscale declined to comment, and Nvidia did not immediately respond to a request for comment. The Information previously reported on the discussions between Nvidia and Nscale, stating that the credit guarantee could be worth several billion dollars.

Earlier this month, Nvidia CEO Jensen Huang stated that he had asked the UK company to "work day and night" to overcome challenges in bringing part of its data center operations online.

In recent weeks, the overall market for "new cloud" AI computing infrastructure technology companies—including Nscale, which primarily provides AI computing power—has remained volatile. Another Nvidia-invested new cloud company, Australia’s Firmus Grid, abandoned its $30 billion IPO plan in Australia; South Korea’s Elice also experienced a difficult debut day in its listing.

Previous reports indicated that Nscale had discussed a second $45 billion deal with Anthropic, involving a second building within the West Virginia facility.

For investors focused on Nvidia’s long-term bull market prospects, the potential for the chip giant to guarantee Nscale holds significance both in stabilizing the supply chain and adding risk factors investors must consider. In the short term, if credit support is implemented, it would help maintain GPU and supporting equipment procurement; in the mid-to-long term, valuation support depends on whether the projects can be delivered on schedule, whether customers can continue to pay, and whether the guarantee exposure matches the cash strength of vendors.

If, in the future, new cloud players like CoreWeave face financing obstacles, upstream participants will inevitably be forced to take on more credit responsibility. Should terminal cash returns remain insufficient, pressure may further affect core suppliers. Nvidia’s proposal to provide guarantees highlights how leading companies in the AI computing power industry chain are actively patching significant gaps in AI financing, but one should not simply conclude from guarantee news that the upward trend of AI financing costs has been reversed, or that the tendency to pessimistically price in an AI bubble has been broken.

From Selling Chips to Providing Credit Support: Nvidia Tries to Unblock Financing Bottlenecks for AI Orders

Nvidia is discussing providing a credit guarantee for an Nscale data center in West Virginia intended for use by Anthropic, focusing on covering power and construction costs. The core significance of this arrangement is leveraging the credit strength of chip suppliers to help computing power operators overcome financing hurdles before projects are completed, powered up, and put into operation. Against the backdrop of new cloud companies facing credit pressure and cloud giants’ cash flows being strictly scrutinized by investors, the AI industry chain faces a more concrete problem: with demand and contracts already in place, who will shoulder the upfront capital and project risks needed to turn these orders into deliverable computing power?

The transaction described in the report is still under discussion; the guarantee amount and specific terms have yet to be disclosed. The multi-billion-dollar credit support cited by The Information is not equivalent to the second $45 billion computing deal being discussed between Nscale and Anthropic; the latter is the potential commercial contract size, not the guarantee amount from Nvidia. The controversy over Nscale’s $30 billion IPO valuation goal also indicates that its expansion plans are simultaneously facing challenges in equity pricing and project financing.

The latest financial reports from cloud giants help explain why the market has begun to raise financing requirements. In Q3, Google Cloud revenue rose 103% year-over-year to $30.709 billion with operating profit at $11.121 billion, but Alphabet Group’s free cash flow for the quarter was negative $9.065 billion; Amazon AWS revenue grew 39.3% to about $46 billion, operating profit around $17.8 billion, but Amazon's trailing twelve-month free cash flow was negative $42.4 billion. Growth and profitability in the cloud business are real, as is the cash consumption for expansion—while group-level negative free cash flow does not directly equate to operating losses in cloud activities, it does show that the pace of construction has outstripped the concurrent internal cash generation, necessitating additional funding.

The market is repricing around this funding gap. On October 29, Alphabet Class A shares closed down 12.6%, the Philadelphia Semiconductor Index dropped about 1.4%, ending four consecutive days of gains; after Amazon’s earnings release, its after-hours share price fell as much as 7%. Beyond cash consumption, Alphabet’s advertising results and Amazon’s overall performance and guidance also influenced trading outcomes. Thus, an accurate assessment is that strong cloud growth alone is no longer sufficient to support valuations—investors require that growth, capital spending, and cash returns all withstand simultaneous scrutiny.

Pressure in the credit market is even more fragmented, as AI infrastructure-related credit debt yields and CDS spreads continue to surge. CoreWeave’s 2031 bonds reached a yield of 15.76% on October 22; its five-year CDS, although narrowing to 1,041.4 basis points on October 27, remains elevated. Using a simplified model with a 40% recovery rate and constant risk intensity, this spread corresponds to a five-year cumulative risk-neutral implied default probability of about 58%, including a risk premium and not to be taken as an actual default prediction. Meanwhile, on October 29, Alphabet’s five-year CDS rose by 5.3 basis points to 84 bps, with the cost for default protection on Meta and Broadcom also increasing that day. The market is not suggesting these companies have identical risks, but is collectively asking: when will ever-expanding AI investment reduce dependence on external financing?

Guarantees Can Protect Orders, but Also Accelerate the Transfer of Customer Risk to AI Chip Suppliers

From a project financing perspective, the role of credit guarantees is to provide an added layer of repayment protection for creditors or other transaction counterparties within a stipulated range. Should Nvidia ultimately provide a legally binding guarantee, relevant financiers may lower their reliance on Nscale’s standalone creditworthiness, improving access to financing, interest rates, or collateral requirements. However, a guarantee does not equate to a cash injection, nor does it mean that Nvidia has already repaid Nscale’s debt; the final risk assumed depends on guarantee caps, scope of coverage, duration, trigger conditions, and recourse arrangements.

The support is focused on power and construction for clear industry logic. GPU orders can only be converted into ongoing, billable computing services following the completion of power supply, cooling, networking, and data center facilities. For large training and inference clusters running AI models intensively, lacking any critical infrastructure could render costly equipment unable to deliver expected output. Construction delays may also prolong the time lag between interest expenses and revenue initiation. Therefore, in helping clients resolve construction financing needs, Nvidia is effectively increasing the likelihood of its chip orders being converted into operational clusters, customer revenue, and subsequent purchases.

Nvidia has previously implemented similar arrangements. Its Q2 FY2027 10-Q disclosed a maximum total exposure for related guarantees at $10.85 billion, including the SB Energy arrangement signed in August, with $10.5 billion allocated to SB Energy and $350 million to other AI cloud project land, power, and main construction guarantees. The SB Energy arrangement takes effect upon meeting conditions by project phase and triggers payment upon specified tenant defaults and related events. This figure represents the contractual maximum potential exposure, not cash already paid or expected losses, and is not the guarantee amount in the current Nscale project.

For investors focused on Nvidia’s bullish prospects, the key shift is the deepening association between returns and risks. Chip suppliers previously mainly assumed product competition, customer procurement, and payment risks; with increased equity investment and project guarantees, they may also be exposed to client construction delays, financing struggles, and losses arising from contract performance issues. Should the same cycle of AI spending slow down both chip sales and customer repayment abilities, suppliers could face the dual impact of weaker revenue and increased guarantee risk.