The purveyors of artificial intelligence talk nonstop about breakthroughs, partnerships and billion-dollar infrastructure bets. But beneath the noise, the dominant mood among users of enterprise AI isn’t excitement. It’s unease.
“I see a lot of fear and confusion right now,” Lynn Comp, vice president and global head of sales for the AI Center of Excellence at Intel, told The Forecast in an interview at .NEXT 2026.
According to Comp, the current AI cycle is creating uncertainty that spans the entire technology sector, from infrastructure teams to enterprise users. What’s more, the transition is more dynamic and unsettling than it was during previous technology waves, including the global rollout of the internet and the mass adoption of mobile devices.
The underlying stress is not abstract, Comp said. Rather, it centers on practical questions: What is artificial intelligence? How is it going to affect me? Should I be using it? How should I be using it?
This anxiety is a powerful undercurrent that will shape the future of AI in business far more than keynote rhetoric or product launches will, Comp said.
AI is an Overwhelming Opportunity
Comp’s 30-year career bridges complex hardware engineering with scalable cloud and business strategies. At Intel, she has a front-row seat to how enterprises are actively adopting and scaling artificial intelligence and as a board member of NeuReality she helps guide the innovative AI software and hardware maker. She understands the silicon and cloud architectures that enable AI. Her experience allows her to see and articulate the technical and business challenges that new technologies bring. The rapid rise of AI is something she reckons with intense curiosity, concern and optimism.
“Don’t be scared, just stay curious,” she told Tech Area. “Do everything in your power to be the positive influence on every collaboration you engage in.”
In a May 2026 Forbes article, Comp described AI as a polarizing technology with very little middle ground when it comes to the positives, negatives and the fear of change driven by it.
“As we navigate the evolving landscape shaped by AI, it’s clear that our most valuable assets are empathy, adaptability and the curiosity to truly care about others’ needs and perspectives,” Comp wrote. “While technology keeps advancing, our success in business development hinges on our willingness to see the world from others’ perspectives and to adjust in the moment as we continue learning through engagements with both humans and AI.
One reason for companies’ AI anxiety is that organizations have realized how difficult it is to deploy AI on their own, Comp told The Forecast. She said the future of AI will hinge on partnerships because large-scale AI infrastructure is too demanding for most enterprises to build independently.
Training advanced AI models often requires hyperscaler-class systems, along with enough power, cooling, floor space and network capacity to support them, Comp said. As a result, companies quickly run into questions about whether they can actually deploy AI infrastructure in their own environments.
“Do I have the power? Do I have the cooling? Do I have the space?” Comp asked. "You don't have any choice but to work together.”
Comp suggested that this pragmatic need to collaborate is reshaping long-standing assumptions across the data center stack. Hyperconverged infrastructure vendors are working more closely with storage partners, for instance, while AI, cloud and distributed infrastructure are forcing a level of interdependence that’s breaking down older category boundaries.
“It's not just compute capacity that’s constrained,” Comp observed. “It’s also memory capacity.”
It’s Still Early Days for AI
For all the bluster about AI’s advances and emerging capabilities, Comp contends that AI is still early in its optimization cycle. Therefore, one should not mistake today’s massive systems for the final form of AI infrastructure. Instead, they’re closer to prototypes of a future that will eventually become smaller, cheaper and more targeted, ventured Comp, who said current AI systems resemble the early hardware demonstrations of mobile infrastructure, when the technology worked but only in oversized, highly configurable forms.
“When mobile phones and mobile telephony first came out, the systems that you were carrying were the size of a box,” Comp said. “They were suitcases.”
For AI technology to become truly pervasive, it will need to move through a similar process of refinement that includes heterogeneous architectures, more specialized designs and, in some cases, a shift from programmable systems toward hardened logic and ASICs, Comp argued.
She frames modern AI infrastructure as a balancing act among compute, data proximity and bandwidth: Whenever one side gets overbuilt, the others become choke points. Raw processor investment alone does not guarantee useful performance if AI accelerators sit idle because networking could not keep up.
If memory capacity is the constraint, autonomous agents can burn through token budgets quickly enough to become a serious operating expense, cautioned Comp, who said multiple agentic runs can end up burning tokens that cost companies as much as a full-time employee. When that happens, optimization stops being a technical preference and becomes a business requirement instead.
AI Governance is Harder Than Engineering
Infrastructure can be solved with engineering and cost discipline, but governance is harder, Comp said. AI agents are the next major fault line, she explained, especially once they begin operating with credentials, privileges, purchasing authority and access to sensitive systems.
“When something needs your user ID credentials, admin privileges, your credit cards, your calendar, you’re taking a really interesting risk,” Comp said.
At this point, the issue is no longer just model safety in the abstract, Comp continued; it’s whether organizations know how to define identity, authority and accountability for autonomous software actors.
Deeper questions around ownership and control remain: What happens when an employee creates an agent that materially boosts productivity? Does that agent belong to the worker or to the company? Do employees take their agents with them when they leave?
AI governance must move from policy slides into enforceable technical and legal structures, Comp argued, pointing to unsettled questions around privilege, copyright and acceptable use, which is not a minor compliance gap but a foundational reset.
“We are going to have to rethink things as dramatically as we did when people started getting connected with the internet,” Comp said.
Practical AI Isn’t About Spectacle
Comp pushes back at the idea that AI has come out of nowhere. Many useful forms of AI have been deployed for many years, even if they weren’t labeled as such, she said, citing media analytics, image inference and recommendation engines as examples of AI systems that have long delivered practical value in narrow domains.
“It's not new like the way that people think of it as being new,” said Comp, who believes practical use cases trump generalized hype. For that reason, she thinks the technical path forward for AI is real but depends on realistic infrastructure choices, narrower use cases, tighter optimization and much more serious thinking about governance.
If she’s right, organizations will get the best outcomes with AI when they constrain the task, understand what they need and design around that requirement instead of expecting one system to do everything.
“If you know what you’re doing and you know what you need and you can actually tighten it up, you end up with some really creative solutions,” Comp said.
Gary Hilson has more than 20 years of experience writing about B2B enterprise technology and the issues affecting IT decisions makers. His work has appeared in many industry publications, including EE Times, Fierce Electronics, Embedded.com, Network Computing, EBN Online, Computing Canada, Channel Daily News and Course Compare.
Ken Kaplan contributed to this story. He is Editor in Chief for The Forecast by Nutanix. Find him on X @kenekaplan and LinkedIn.
© 2026 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.
Related Articles
