Being on the ground in Las Vegas for Dell Technologies World 2026, one thing became clear almost immediately. The conversation around AI has fundamentally shifted.
We've stopped talking about what's possible and started focusing on what's practical.
As Michael Dell and Jensen Huang framed it from the stage, we've moved past "novel AI" into the era of useful AI, and that set the tone for everything that followed.
Expectations have changed fast
We've gone from talking about 20 to 30% productivity improvements to 20 to 30 times gains with agentic AI. But the number that should grab every partner's attention is token consumption, projected to grow 3,400% by 2030.
Customers aren't experimenting anymore. They're rethinking how work gets done, and they're burning through compute at a rate that breaks old assumptions about cost.
The token crunch is the real story
Here's what that 3,400% figure actually means in practice.
Tokens have become the new unit of cost, and at scale they get expensive fast. An agentic workflow that looks cheap as a pilot can become punishing once it runs across an entire business, thousands of times a day. The economics that made sense for the occasional cloud inference call start to fall apart when AI becomes ambient.
This is the context most hybrid AI conversations are missing. When Michael Dell said hybrid AI is a competitive advantage rather than a compromise, the advantage is financial.
Running every token through public cloud inference is getting harder to justify. Hybrid lets organisations put predictable, high-volume workloads on-prem or at the edge where the per-token economics are controllable, and reserve cloud for the bursts and the experiments. So, the balance across cloud, on-prem and edge has become a question of cost as much as performance or control. At real scale, it's the difference between a workload that pays for itself and one that quietly bleeds you.
For partners, that's where the strategic value sits. Designing hybrid AI environments takes an understanding of workloads, data movement, compliance, and the cost curve of running inference at volume. That's exactly where Dicker Data supports partners through AI Accelerate, translating vendor innovation into deployment strategies that are actually affordable to run.
Infrastructure is back in focus
Infrastructure is no longer sitting in the background. AI workloads are changing how environments get designed, from compute to power density to storage performance, and the token economics above are a big reason on-prem capacity suddenly looks attractive again.
For partners, that's a strong opening to re-engage customers around modernisation, particularly where existing environments aren't fit for AI workloads. The conversation shifts from transactional upgrades to strategic transformation.
A partner-led moment
The big takeaway from Dell Technologies World is that we're entering the phase where AI becomes real, and partners who are ready to deliver will lead it.
The technology is there, demand is building, and customers want guidance on what comes next, whether that's an infrastructure refresh, evolving workloads, or getting ready for AI at scale.
For us at Dicker Data, being on the ground reinforced how important that role is. Through AI Accelerate, our focus is helping partners understand the opportunity and build the capability to capture it. This is a fundamental shift in how businesses operate, and the partners who move now will be the ones who lead it.
Get in touch with our local Dicker Data team to find out more.
