The fix for soaring AI cloud bills exists — so why won’t we trust it?
To hear Yasmin Rajabi, chief operating officer at CloudBolt, tell it, there’s an imbalance in how we view automation. We’re happy to automate decisions that result in more productivity and processes — but what about when it comes to turning the dial to the left? For some reason, there’s hesitation there.
“Trust is super-high when it comes to traditional automation, but there’s still a lot of caution when it comes to right-sizing,” Rajabi tells The New Stack. “The same engineers who are deploying multiple times a day through CI/CD aren’t questioning [automation] anymore, but when it comes to delegating right-sizing to the machine, the bar to earn that trust is much higher.”
The data reveal why this imbalance might exist: When faced with the pressure to remain always-on, a higher cloud bill from over-provisioning seems worth the cost. But now that GPU-heavy AI workloads have sent cloud bills soaring, right-sizing automated processes has become a priority for 89% of organizations, according to the March 2026 CloudBolt Research Report.
And yet, 71% of Kubernetes engineers respond that they still require human review for resource optimization, with only 27% allowing CPU and memory changes to be auto-applied. So while the data shows it’s a priority, that motivation hasn’t shown up in the workflows.
“When it comes to delegating right-sizing to the machine, the bar to earn that trust is much higher.”
The New Stack will sit down with Rajabi and Reid Vandewiele, product lead at StormForge, at 9 a.m. Pacific/5 p.m. BST on Wednesday, June 24 to discuss the urgency of this right-sizing gap — especially when it comes to Kubernetes workloads for AI.
Join us live to not only learn how to measure your organization’s automation maturity, but to develop this trust over time, with strategic CPU throttling, out-of-memory (OOM) behavior, and, of course, rollback patterns.
Register to join this conversation
Right-sizing is a multi-dimensional problem, Rajabi explains, spanning increasingly complex workloads in increasingly complex environments, so that when something goes wrong, it feels almost impossible to reverse. For this automation to work, this trust has to be built not just across teams but scaled across your organization.
“It takes a long time to build up trust in an automation solution, and it’s very fast to eliminate or significantly undermine that trust,” Rajabi warns. “It takes one production incident to take an application team from being willing to entertain automated resourcing to absolutely not, ‘not on my application, we’re special’.”
Join us June 24 to learn how to gain insight into how much your AI workloads actually cost, and to adopt a plan that takes the guesswork out of your provisioning.
