2 min read
Add as a preferred source on Google

Vultr Says Enterprise AI Agents Need Cloud Services Beyond GPUs

Kevin Cochrane says production agents also need compute, storage, networking and secure environments to work with business systems.

A glass enclosure isolates a compact compute unit in a data center corridor / TokenPost.ai
A glass enclosure isolates a compact compute unit in a data center corridor / TokenPost.ai

As enterprises put AI agents into production, Vultr says the systems need infrastructure for tasks beyond sending requests to graphics processing units (GPUs), including running code and accessing company systems.

Kevin Cochrane, Vultr’s chief marketing officer, said businesses need a broader mix of cloud services. “They need a full cloud platform,” Cochrane said. “They need cloud compute, specialized AI compute, networking, storage and an open ecosystem.”

Agents can run scripts, build software, query databases and use external tools as well as send requests to AI models. Those tasks can require central processing units (CPUs), storage and networking alongside GPU resources. Cochrane also described secure sandboxes, which isolate agent activity from production systems and sensitive resources, as an emerging need.

Vultr entered the cloud market in 2014. It offers traditional cloud services alongside specialized GPU resources. Cochrane said customers can deploy more than 60 projects from the Cloud Native Computing Foundation through Vultr’s platform.

Regional operations are another part of the infrastructure question. Cochrane described Vultr’s 34 data center regions as independent, with data kept within its region. He said an application in France would continue running if the company’s global control plane and other regions became unavailable.

The shift toward production agents is prompting companies to consider how they govern systems with access to business tools. Cloud providers are extending their offerings beyond GPU capacity to support the broader infrastructure needs of AI applications.

Simon Yoon

Reporter

Simon Yoon reports on blockchain technology for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…