Vultr at Gartner Symposium: Enterprise AI Needs More Than GPUs

Enterprise AI infrastructure has to be built around the applications companies want to run, not only around the accelerators they want to buy, according to Vultr. The cloud provider shared three takeaways from Gartner IT Symposium, covering GPU limits, the continuing role of CPUs and a workload-first approach to deployment.
- Vultr says enterprise AI needs complete cloud infrastructure, not only GPU capacity.
- Gartner's October 2026 announcement said more than $3.5 trillion in AI infrastructure spending could be impaired by the end of 2030.
- Vultr CMO Kevin Cochrane said the CPU never went away, and Vultr sees rising CPU demand from agentic AI.
- A panel moderated by HPE advised starting with the workload and measuring business outcomes.
- Vultr's AMD EPYC-powered VX1 Cloud Compute offers dedicated CPU resources for AI orchestration.
Beyond GPU capacity
Vultr says the rise of neocloud providers and growing investment in AI infrastructure were major topics at the Symposium. As organizations commit more resources to AI, questions about infrastructure economics, long-term value and the ability to turn spending into business outcomes are gaining weight.
Vultr pointed to Gartner's October 2026 announcement that more than $3.5 trillion in AI infrastructure spending could be impaired by the end of 2030. According to Vultr, the announcement added urgency to these conversations and underscored the need to evaluate infrastructure decisions carefully.
For Vultr, the opportunity goes beyond providing GPU as a service (GPUaaS). The company says enterprises need a complete cloud foundation covering compute, accelerated infrastructure, storage, networking and security. It also argues that companies should be able to choose infrastructure based on technical and business requirements, rather than assuming every AI workload needs the same architecture.
CPUs stay relevant as agents spread
The role of CPUs was another recurring theme. Vultr CMO Kevin Cochrane reinforced a simple point in his discussions: the CPU never went away.
Vultr's reasoning is that AI agents do more than training and inference. They coordinate tasks, call tools, retrieve information, interact with existing systems and execute multi-step workflows. These activities place extra demands on CPU resources alongside GPUs. Cochrane said in a recent interview with Fierce Network that Vultr is seeing demand for CPU resources increase as agentic AI workloads become more prevalent.
Vultr links this to its AMD EPYC-powered VX1 Cloud Compute, which it says provides dedicated CPU resources for the application logic and orchestration behind modern AI workloads. The company's conclusion is that organizations should account for the entire workload rather than focusing only on GPUs.
Start with the workload, then measure
Vultr's Piyush Shukla took part in a customer panel moderated by HPE, alongside Sudheer Matta, SVP of HPE Networking, and Laura Lincoln of Grant Thornton. The discussion centered on scaling AI with trust, control and measurable impact.
A key point was to start with the workload and work backward. Before choosing a model or investing in infrastructure, organizations should identify the business problem, the desired outcome and the application's requirements, then select resources and architecture to match. Some workloads need substantial GPU capacity, while others lean on CPU performance, data access, storage or networking. Many production applications need a mix.
The panel also stressed metrics. Deploying an AI application is only the beginning, and organizations need to define how they will measure success and demonstrate business impact. On security, the panel noted that as AI agents gain the ability to act on enterprise systems, security must be part of the infrastructure conversation from the start.
What it means for enterprise buyers
Vultr says its Symposium messaging reflects its ongoing enterprise AI investments, including its Chennai cloud data center region, which expands its global footprint and access to infrastructure for AI and enterprise workloads. It also cited enterprise AI platform developments aimed at letting organizations combine compute, accelerated infrastructure and supporting technologies as needed.
The company also observes that infrastructure purchasing increasingly involves security leaders alongside IT, engineering and business stakeholders, which makes architecture and governance a joint concern. Cochrane used an ultramarathon analogy: enterprises have reached the end of the first marathon, but the journey is far from over.
The priorities Vultr drew from the event are to look beyond GPU capacity, account for the growing role of CPUs, begin infrastructure decisions with the workload and measure success by business outcomes. These are the company's own views, presented from the perspective of a cloud vendor.


