AI Cloud Prices Rise by 25%, Intensifying Platform Competition

AI cloud companies are shifting their focus from GPU rental-based businesses to AI infrastructure platform businesses, driven by recent performance results. The demand for AI computing is outpacing short-term supply, leading to a reevaluation of prices and contract structures.

Nebius and CoreWeave recently revealed the growing demand for AI cloud services and the burden of large-scale capital investments in their earnings reports. The performance of both companies indicates that the market's focus is shifting from merely securing GPUs to platform competition that includes software and operational capabilities.

CoreWeave is expanding its large customer contracts based on its AI cloud business. An analysis reported by O'Daily highlighted that CoreWeave's recent performance exemplifies the rapid expansion phase of the AI cloud service provider (CSP) industry.

On the afternoon of the 13th at 3:58 PM (Korean time), analyst qinbafrank posted an analysis on X, stating that the recent performances of CoreWeave and Nebius demonstrate the rapid expansion phase of the AI cloud service provider industry. However, since this analysis was presented by a media outlet, it is necessary to read it separately from the figures disclosed by each company in their public announcements and shareholder letters.

A common point revealed in the performances of both companies is the shortage of AI computing supply. The analysis indicated that Nebius's average annual revenue from new contracts in the second quarter exceeded $20 million per MW (approximately 2.83 billion KRW), with some projects reaching between $20 million and $25 million per MW.

The price increase is not a uniform trend across all GPU resources. According to the analysis reported by O'Daily, CoreWeave's GPU computing SKU prices rose by about 25% in July, while Nebius's prices for older generation GPUs increased by over 30% compared to the first quarter. The price for short-term emergency capacity was reported to be between $40 million and $50 million per MW (approximately 5.67 billion to 7.09 billion KRW).

The price reevaluation is first appearing in high-value resources that can be deployed immediately, such as short-term capacity, new generation GPUs, large clusters, and production AI inference. In contrast, existing computing resources that are lower in priority or locked in at long-term fixed prices find it difficult to reflect the same level of price increase.

Revenue models are also changing. The analysis explained that Nebius is adopting a lightweight asset model utilizing capital partners, while CoreWeave is reducing GPU investment burdens through long-term contracts and asset-based financing. Long-term contracts and financial structures serve as mechanisms to lower the initial cost burdens associated with data centers and GPU procurement.

However, depreciation and financing costs remain burdensome. The analysis cited by O'Daily noted that both companies are still in the stage of large-scale capital investment, and depreciation and financing costs are pressuring profit margins.

CSP stands for cloud service provider. AI cloud refers to infrastructure that bundles GPUs, networks, storage, deployment tools, and operational automation functions necessary for model training and inference. Until now, the focus has been on hourly GPU rentals, but recent competition has broadened to who can provide a more stable overall environment for customers to operate actual AI services.

Nebius is promoting a platform that encompasses data, model training, and deployment. The analysis explained that Nebius is combining its own AI infrastructure operations and software capabilities with AI data centers built with partner capital.

CoreWeave is also expanding its services in the direction of integrating AI research, inference, and agent execution environments. This shows that the competitive axis among CSPs is shifting from merely securing GPUs to software, networks, storage, and operational automation. Previously, we reported that NVIDIA explained a structure where it supplies the GPUs, CPUs, and networks necessary for training and inference of large-scale AI models through CoreWeave and Nebius as a single platform.

In the domestic market, this trend leads to the question of who operates the computing infrastructure and sets prices rather than the AI semiconductors themselves. The GPU supply shortage is reflected in pricing negotiation power and contract structures, and AI cloud operators are trying to absorb the burden of capital investment through long-term contracts and platform-based services.

A similar trend is also emerging in enterprise AI operating environments. We reported that Spectro Cloud launched a local-first inference platform that runs AI models within companies. This case illustrates that the competition for AI infrastructure is not limited to public cloud GPU rentals but is expanding into private infrastructure, inference operations, and data control areas.

The analysis surrounding the recent performances of CoreWeave and Nebius shows that the demand for AI computing still outstrips supply, while prices, contracts, and platform strategies are changing simultaneously. The next point of observation is how much the speed of new capacity supply and long-term contracts can offset depreciation and financing costs.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.