For more than a decade, microservices and containers were a solution in search of a problem. Engineers admired the architecture from a distance, then went back to their monoliths. Artificial intelligence changed the math.
"Cloud native is rediscovering itself with AI workload," said Deepak Goel, chief technology officer of cloud native at Nutanix, in an interview.
"The properties that cloud native used to advertise: you can break your monolith application into microservices, you can roll out these microservices independently. The same is happening with AI workload."
Goel leads cloud native product and engineering strategy at Nutanix, where he oversees the platforms that package and orchestrate containerized applications across hybrid environments.
When he looks at the trajectory of cloud native technology, he sees a parallel with graphics processing units.
“Nvidia built GPUs for gamers in 1993,” he said. “The hardware idled for years until cryptocurrency and then AI sent demand skyward. Cloud native followed a similar arc, waiting a decade for the workload that would justify its complexity.”
Goel acknowledged that not everyone jumped on the cloud native bandwagon.
“Frankly speaking, nobody wants to poke a sleeping bear…so if a monolith is working fine, they don’t want to go and break it down and have a business impact on it.”
But the growing use of cloud native technologies is irrefutable. According to the 2026 Enterprise Cloud Index from Nutanix, 85% of IT executives say AI is accelerating their adoption of the containers Kubernetes enables, and 94% call cloud native architectures the "gold standard" for deploying modern AI applications at scale.
Breaking the Monolith, Accelerating the Cycle
The appeal is speed. A monolith forces a waterfall rhythm: design, change, test, regress, ship. Break that single binary into 10 services and 10 teams can build, test and release in parallel. That’s why the faster pace of cloud native is so appealing.
"You can roll out a part of your application. Testing becomes more modular. Your application becomes more modular," Goel said.
"Your updates become more continuous. That reduces the time to market."
The Complexity Tax
The flexibility is not free. One binary becomes 10, and every operational task multiplies with it. Observability, security and networking all grow more intricate.
"Earlier, you were securing just one application; now you're securing 10 services," Goel said.
That shift also introduces east-west traffic, the chatter between services that never existed when a single program handled everything internally.
"Your fault observation has increased from one binary to 10 services," he said, explaining that IT teams must suddenly package, deliver, secure, and monitor ten separate entities. “What was one problem has become ten in terms of operational burden.”
A Platform for a Heterogeneous World
Goel's prescription is a unified platform, not a single cloud. Enterprises run Java and Python, Windows and Linux, virtual machines and containers, on premises and in the cloud. Each environment tends to spawn its own team and its own tooling.
"Workloads will find the place where they need to run," Goel said. "But what happens to the organization is they need different toolings, different sets of experts. That increases the operational cost."
Instead, a platform that follows workloads wherever they land, with consistent security, networking and observability interfaces, lets one team manage the sprawl.
"You can manage this whole heterogeneity with a single team," he said.
Pitfalls and the Credit Card Bill
The most common mistake, Goel said, is treating modernization as a cure-all. Teams decompose monoliths so aggressively that the pieces no longer fit together. Others chase an "all cloud" mandate and recoil at the invoice.
"Cloud is built very interestingly,” Goel said. “On surface, it appears that everything is very cheap. But when you put together everything, at the end of the monthly bill is very expensive."
He compared it to paying by credit card: the ease hides the spend. Many organizations then repatriate workloads back on premises.
“It is the same thing as paying by cash or paying by credit card,” he said. “You often tend to spend more when you are paying by credit card, just because it is so easy to do.”
His advice is to define success before starting, decompose selectively and keep a hybrid stance.
"See how moving to the cloud is helping them?” he retorted. “Is it really helping them reduce their cost, or is it just a shiny thing?"
Repurposing Stranded Hardware
Cloud native also answers a hardware problem. Data center admins tell Goel that GPUs become obsolete before they recoup their investment. Enterprises that hedged on AI hardware now hold aging capacity.
"Cloud native is helping reutilize those machines in an effective way," Goel said.
Because it runs on any compute, storage and network, it can expose idle GPUs to the AI workloads that need them.
Cloud native technologies have grown up a lot in the past decade, but the rise of AI is putting its real potential to good use. But Goal warns that without a unified, platform centric architecture to manage the resulting complexity, enterprises risk collapsing under the weight of the very systems they built to differentiate themselves.
Related:
How AI Dramatically Disrupts IT Operations Unlike Anything Before
In the AI Era, Cloud Native Is No Longer Optional
Kubernetes Everywhere - Virtualized, Bare Metal and Built for AI
Why Kubernetes Efficiency is About Choice, Not Compromise
How Cloud Native AI is Driving Enterprise Transformation
AI and Cloud Native Spark Explosion of New Apps
Kubernetes Grew Up Just in Time for AI, Says Analyst Steven Dickens
Study Shows Big Uptake of Enterprise AI and Cloud Native Technologies
Jack Gibson is a contributing writer for The Forecast by Nutanix. (Jack Gibson is a 2026 Corporate Communications Intern at Nutanix. And a rising senior at Loyola Marymount University studying Marketing. Find him on LinkedIn.)
© 2026 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.
Related Articles
