VentureBeat survey reveals 79% of enterprises report idle GPUs

GPT: GPT denotes OpenAI's frontier model lineup that powers many enterprise and consumer AI interactions via APIs. It is referenced in the reporting as a primary example of models that remain closed and require separate infrastructure, contributing to the observed split in AI spending. Claude: Claude refers to Anthropic's frontier model family used for advanced reasoning and agentic workflows through API access. The news highlights Claude as one of the preferred high-capability options driving token-based spending that does not leverage existing enterprise GPU resources. Google: Google provides Google Cloud as a platform for AI compute and model hosting, including integration with Anthropic and OpenAI offerings. The news references Google among cloud giants whose capacity and model services are central to enterprise AI spending debates. Its role highlights how frontier models reach customers through established cloud channels. Nvidia: Nvidia develops and supplies GPUs that power AI workloads across training and inference. It is positioned in the news as one of the primary providers of AI compute hardware facing questions about overcapacity from enterprise customers. Recent reports tie its hardware sales directly to the mismatch between reserved GPU resources and actual utilization patterns. OpenAI: OpenAI develops the GPT series of frontier models delivered via APIs and now expanding to additional cloud platforms. In this reporting, OpenAI's models exemplify the second category of AI demand that cannot run on customer-controlled GPUs and sustains ongoing API expenditures. Recent announcements include broader cloud availability for its offerings. Anthropic: Anthropic builds and maintains the Claude family of frontier AI models accessed primarily through APIs. The news positions Anthropic's models as a core driver of per-token spending that remains separate from enterprise GPU capacity reservations. Its offerings are integrated into major cloud platforms and continue to attract usage even as hardware sits idle. Microsoft: Microsoft operates Azure, a leading cloud platform that hosts and serves AI models from partners including Anthropic and OpenAI. In the context of this news, Microsoft is cited among major cloud providers where enterprises access frontier models via APIs rather than running them on their own idle GPUs. Its infrastructure offerings are part of the broader discussion on enterprise AI platform switching decisions. Amazon Web Services: Amazon Web Services runs a major cloud platform that delivers AI compute capacity and hosts models from providers such as Anthropic and OpenAI. It appears in the news as one of the key sellers of AI infrastructure that enterprises evaluate when considering new or replacement platforms despite existing idle GPUs. Praveen Neppalli Naga: Praveen Neppalli Naga serves as chief technology officer at Uber and has commented publicly on enterprise AI budget realities. His August remarks in the news describe the end of an initial phase of heavy model usage spending, illustrating how companies are confronting rising AI costs separate from hardware utilization. Enterprise AI Strategy: Demand for frontier models from providers like Anthropic and OpenAI continues through APIs even as many companies maintain reserved but underused GPU capacity. Infrastructure Planning: Companies running AI workloads in production are actively evaluating new or additional compute platforms within the coming year regardless of current GPU utilization levels. Cloud Platform Integration: Major cloud providers including Microsoft, Google, and Amazon Web Services now host and serve frontier models from OpenAI and Anthropic, expanding enterprise access options beyond owned hardware.