While the GPU-as-a-Service (GPUaaS) market, which provides graphics processing units as computing resources, continues to grow rapidly, three out of four companies in the sector are expected to either shift their business models or exit the market entirely.
Global IT advisory firm Gartner, in its recent report *GPUaaS Is Ready to Support AI—But Are You?*, advised that enterprises should not select partners based solely on GPU pricing or availability.
GPUaaS allows businesses to rent AI-grade GPUs via the cloud instead of purchasing them outright. Gartner emphasized that companies must conduct comprehensive evaluations of providers, including financial stability, service scope, technical support systems, and partner ecosystems, before adoption.
The report predicts that by 2030, 75% of current GPUaaS providers will either △Pivot to general-purpose cloud services focused on non-AI workloads △Be acquired by hyperscalers △Exit the market entirely.
This outlook stems from the market's high capital requirements and short hardware replacement cycles. Providers that merely supply infrastructure without differentiated value or long-term strategies are likely to struggle. Many still lack the maturity of hyperscalers in service capabilities and technical support.
Gartner noted, “Early GPUaaS providers competed on immediate access to AI-optimized GPUs and massive power capacity. However, hyperscalers now match or exceed these offerings while integrating enterprise data, workloads, and proprietary AI accelerators and models—narrowing the niche for standalone GPUaaS players.”
Financial instability is another critical risk. Many pure-play GPUaaS providers, such as CoreWeave, Crusoe, Lambda Labs, Navius, and Nscaling, remain unprofitable and rely on external investments or debt. Overdependence on a few large AI model developers for revenue and utilization rates further undermines long-term stability.
Excessive reliance on NVIDIA is also a structural weakness. While hyperscalers like AWS (Trainium), Google (TPU), and Microsoft (Maia) diversify supply chains with in-house AI accelerators, most GPUaaS providers remain tied to NVIDIA and AMD for hardware and financial support.
Gartner stressed, “Scrutinizing providers' paid customer bases, profitability, and break-even timelines is essential. Enterprises must also prepare exit scenarios in case providers fail or suspend services amid volatile market conditions.”
