CoreWeave warns: It will face difficulties if forced to abandon Nvidia chips

AI cloud computing provider CoreWeave has, in a rare move, disclosed a potential risk in its latest regulatory filing: if market demand shifts and it is forced to abandon its reliance on NVIDIA chips, the company will face significant time and financial pressures.

In its quarterly financial filing submitted on Wednesday, CoreWeave wrote that changes in customer demand could require the company to invest “time, capital, and resources” to source alternatives to NVIDIA GPUs. It warned that this could ultimately impact its ability to provide the necessary products to its customers. This is the first time the company has formally listed “de-NVIDIA-ization” as a potential business risk in a regulatory filing.

Meanwhile, CoreWeave’s earnings outlook exceeded market expectations, with an order backlog reaching 104 billion dollars by the end of the second quarter. The news boosted shares, which jumped 19% on Wednesday to close at $107.73.

Currently, NVIDIA holds about 10% equity in CoreWeave and provides all GPU computing power for all its customers. This highly concentrated supply structure means that any supply chain switch would have wide-reaching implications.

“De-NVIDIA-ization” cited as a risk factor for the first time

This warning appears in the “Risk Factors” section of CoreWeave’s quarterly filing—a part typically used to disclose hypothetical scenarios that could impact the business. CoreWeave declined to comment further on the language of the document.

The introduction of this new wording suggests that CoreWeave has internally begun to consider scenarios where NVIDIA’s dominance in AI chips weakens. This marks a subtle shift from the company’s previous public position: in December last year, CoreWeave executive Nick Robbins stated that customer demand was “overwhelmingly” still focused on NVIDIA technology and that the company would only consider strategy adjustments if there was a large-scale, systemic change in customer preferences.

Major clients accelerating in-house chip development

The backdrop for CoreWeave’s risk disclosure is the accelerated push by its core client base toward developing their own chips. One of CoreWeave’s largest clients, OpenAI, has already released its first AI accelerator chip custom-built with Broadcom. More broadly, several major AI computing power purchasers are working to reduce their reliance on NVIDIA chips, instead developing their own silicon solutions.

Nevertheless, NVIDIA remains the undisputed leader in the AI chip sector. CoreWeave, as a representative of the neocloud service provider cohort, has a business model highly dependent on NVIDIA’s GPU supply—the latter currently supplies all GPU resources to all CoreWeave customers.

Strong results provide short-term support

Aside from risk disclosures, CoreWeave’s results this quarter remain robust. On Tuesday, the company’s revenue outlook surpassed analyst expectations, and its order backlog reached 104 billion dollars, offering the market considerable business visibility. Buoyed by this, CoreWeave’s share price surged 19% on Wednesday.

Strong order numbers indicate that, for the foreseeable future, market demand for NVIDIA-driven AI computing power remains resilient. However, CoreWeave’s proactive addition of supply chain risk to its formal filings also sends a signal to investors: as the AI chip ecosystem evolves, this cloud computing provider, closely tied to NVIDIA, is preparing for a more complex competitive landscape.