Dell and Intel tackle enterprise AI's token budget crisis with on-premises alternatives

The economics of enterprise AI deployment are forcing a fundamental shift in strategy. As agentic workloads proliferate, organizations are exhausting their cloud token budgets far faster than anticipated, creating pressure to reconsider the balance between cloud-based frontier models and on-premises infrastructure paired with open-weight alternatives.

Helen O'Sullivan, AI business development manager and solutions specialist at Dell Technologies Inc., observed this pattern firsthand among her customer base. "A lot of our customers I support in the Bay Area, they're dealing with this real immediate struggle of running through their token budget in a matter of weeks that they had allotted for months," she said. "You still use those frontier models, but you use the open source and infrastructure on-prem to at least eliminate some of the token stress and token maxing going on right now."

O'Sullivan and Tim Wood, sales director for the Northwest region at Intel Corp., discussed these dynamics during an exclusive conversation at Dell's AI Leadership Symposium, broadcast on theCUBE, SiliconANGLE Media's livestreaming studio. Their remarks highlighted how cost constraints are reshaping where and how enterprises deploy AI workloads.

Rethinking hardware allocation for agentic workloads

The shift toward agentic AI is altering the hardware requirements enterprises face. While many organizations remain focused on graphics processing units, the computational demands of agentic systems are creating a different processor mix than traditional AI deployments.

O'Sullivan emphasized that strategic alignment matters more than raw capability. "You can no longer do AI just for the sake of doing AI. You need to have a goal in mind that is in line with the corporation's goals," she said.

Intel's approach to agentic AI spans orchestration, inference and edge computing. O'Sullivan's team anticipated this hardware rebalancing months ago, though adoption remains uneven. "We saw the rebalance and the heavier utilization on the CPU when it comes to agents. Customers are getting there," she said. "They're not quite there yet from the scope I'm seeing."

Evaluating existing infrastructure before deployment

Both companies recommend a consultative approach that assesses existing capital investments before committing to new infrastructure. Testing open-weight models in controlled environments allows enterprises to understand their actual workload requirements without immediate production risk.

Wood noted that many organizations already possess underutilized CPU capacity. "A lot of enterprises already have CPUs in their data center. There may be existing resources and headroom in the infrastructure they already have that's capable of doing some of these next-generation apps and workloads," he said.

https://www.youtube.com/embed/Xg7XBk0Nweo?feature=oembed

O'Sullivan stressed that effective AI deployment requires examining the entire technology stack, not just accelerator performance. "A lot of [customers] just see the [graphics processing unit] and [ask], 'How fast can I get the GPU?' or Vera Rubin or what have you," she said. "But it's about the data."