CoreWeave Stock Jumps After Launching AI-Optimized Object Storage

Shares of CoreWeave (ticker: CRWV) rose about 2% following its announcement of a new AI-native object storage service, signaling investor confidence in its infrastructure push. The company aims to cut storage costs for AI workloads by more than 75% while eliminating egress, request, and transaction fees. 

This move extends CoreWeave’s ambitions beyond compute and GPU clusters into the data layer of AI systems. The storage product is fully managed and designed to allow a single dataset to be accessed across regions, clouds, or on premises, without duplicating data or paying extra fees. The modest stock reaction underscores cautious optimism from the market, as analysts weigh the potential of CoreWeave challenging major cloud incumbents.

What Is CoreWeave’s AI Object Storage?

CoreWeave’s new AI Object Storage is a managed, S3-compatible object storage service built specifically for AI training and inference. Unlike general-purpose object storage, it integrates tightly with GPU compute nodes, aiming to minimize latency and maximize throughput. One of its core elements is the Local Object Transport Accelerator (LOTA), a proxy deployed on each GPU node that caches frequently used data locally and pre-stages objects on NVMe drives, reducing traffic across the network.

In benchmarks, the system supports 2 GB/s per GPU throughput and is engineered to scale across large clusters of GPUs. The architecture enables a single dataset to be accessed seamlessly across regions, clouds, or on premises, removing the need for dataset duplication.

On the cost side, CoreWeave introduces three usage tiers (hot, warm, cold) and promises no egress, request, or transaction fees. The company claims this model can reduce storage costs by over 75% for typical AI workloads.

Additional features include durability targets, encryption at rest and in transit, role-based access control, and observability via dashboards. The system is designed to scale to trillions of objects and exabytes of data, preserving throughput and performance as workloads grow.

One recent update cites throughput potential up to 7 GB/s per GPU in certain configurations as the technology matures. CoreWeave also highlights that it achieves cross-cloud and cross-region access without requiring users to replicate data. 

In short, this product is a strategic bet that high-performance, AI-aware storage will become a key differentiator for infrastructure providers. Its success hinges on delivering reliable throughput under scale and convincing AI teams to adopt a new storage paradigm.

Strategic Implications & Competitive Context

With this move, CoreWeave is expanding from pure compute infrastructure into the data stack, potentially making its platform more “sticky” for large AI users. By owning both compute and storage optimized for AI workloads, the company can offer tighter integration and reduce data movement friction.

Its pricing model, especially the elimination of egress and transaction fees, directly challenges the prevailing cost structures of major clouds like AWS, Google Cloud, and Azure, which often charge users to move data out of storage buckets. If CoreWeave’s model proves viable at scale, it may pressure those incumbents to rethink their storage pricing strategies.

CoreWeave’s strategy is bolstered by existing momentum. The company recently secured a major $14.2 billion multi-year AI infrastructure deal with Meta, which sent its stock higher. It is also acquiring AI tools firms like Weights & Biases, tightening its control over both compute and workflow tooling.

Another recent development is its partnership with Poolside, in which CoreWeave will provide more than 40,000 Nvidia GPUs and act as anchor tenant for a large West Texas AI data center campus. This underscores that the company is aggressively scaling its physical infrastructure footprint alongside its software stack.

From the competitive side, large cloud players could respond by slashing storage egress fees, bundling storage with compute, or accelerating investment in performance-optimized storage technologies. Analysts and potential customers will closely watch whether CoreWeave can deliver consistent performance under large multi-region workloads.

If the AI industry’s growth continues, demand for AI-tailored data infrastructure may intensify. CoreWeave’s gamble is that customers will value end-to-end infrastructure optimized for AI more than piecing together compute on one platform and storage elsewhere. The success of this product could either consolidate its position or provoke fierce competition from incumbent cloud providers.

Risks, Challenges & Analyst Views

CoreWeave faces several significant risks that could derail its storage and infrastructure ambitions. One is its high level of debt. In Q2 2025, operating expenses jumped sharply and the company reported a net loss of $290.5 million, far above analyst expectations, raising doubts about its ability to control costs. Its 2025 capital expenditure guidance, $20–23 billion, up from about $8.7 billion in 2024, also adds pressure on margins.

Dependence on a few major clients is another concern. Microsoft, for example, accounted for a large share of its early revenues, making CoreWeave vulnerable if any major client reduces business or builds internal alternatives. Some analysts at HSBC downgraded the stock, citing low differentiation and overreliance on Microsoft and Nvidia.

Analyst sentiment is mixed. Argus initiated a “Buy” rating, forecasting 363% growth in 2025 revenue versus pro forma 2024, though it flagged execution risk. Others remain cautious, some rate the stock a “Hold” or “Reduce,” warning that its valuation is aggressive and that debt and customer concentration leave little room for error.

A bearish view also comes from short-seller Kerrisdale, which argued the business model is “debt-heavy” and lacks meaningful competitive moats. Such critiques underscore that the upside is coupled with substantial financial risk.

Conclusion

Several triggers could validate or undermine CoreWeave’s AI storage ambitions over the next quarters. One critical metric is customer uptake, whether major AI labs or enterprises adopt the storage service at scale. If early customers publish benchmarks showing consistent throughput and cost savings, confidence may rise.

Another factor is how incumbents respond. If AWS, Azure or Google Cloud reduce or eliminate egress fees or introduce AI-optimized object storage, CoreWeave’s pricing advantage may shrink. Tracking changes in cloud pricing policies will be important.

The pending Core Scientific acquisition, expected to close in Q4, could help by reducing lease costs and giving CoreWeave more control over its physical infrastructure. But some large shareholders of Core Scientific oppose the deal’s terms. 

Nvidia’s upcoming earnings also matter. Because CoreWeave depends heavily on Nvidia’s GPU roadmap and demand, disappointing results from Nvidia might pressure sentiment toward compute and infrastructure plays like CoreWeave.

If CoreWeave can execute well, and maintain financial discipline, its storage product could become a differentiator. But until then, the company must balance aggressive growth with the reality of capital intensity and performance demands.