Last week, IBM CEO Arvind Krishna warned investors that the company's infrastructure division revenue declined more than expected year over year, drawing market attention. Just one week before the Q2 2026 earnings release, Krishna said that due to soaring component prices, customers have shifted capital investments from mainframes to servers, storage, and memory hardware.

Despite record growth in IBM's distributed infrastructure division, which covers hybrid cloud, storage, and edge computing, the company's stock price fell on the news.

The mainframe platform's struggles are nothing new. For over a decade, cloud computing was once seen as an existential threat to the platform, until hyperscale cloud providers acknowledged that hybrid strategies have become the norm for enterprises. Recently, large language model assistants designed to refactor legacy COBOL code have sent mixed signals to the mainframe community.

"We went through a little bubble where the narrative was that AI would help translate COBOL, so all mainframes would disappear," Brian Klingbeil, chief strategy officer at managed service provider Ensono, told Channel Dive. "Then people started realizing that AI can also shore up the platform's weaknesses, making it an ideal place for storage. Now, you can modernize the mainframe and make it more agile."

Klingbeil compared the mainframe platform to Apple's iPhone. "It's purpose-built hardware with deep end-to-end integration, tightly coupled with IBM's z/OS operating system," he said. "The iPhone is also an excellent piece of high-end hardware, with great performance, high security, and deep integration with iOS."

Ensono has built a solid business around hybrid cloud management. When the MSP celebrated its tenth anniversary in January, it claimed annual revenue of $1.1 billion and more than 250 customers. Klingbeil estimates that recurring revenue from service offerings accounts for about 85% of Ensono's business, with the remaining 15% coming from project-based revenue.

"When large enterprises have major transformation visions, we get many calls from large systems integrators who want to win the project but lack mainframe capabilities," Klingbeil said.

Mainframe modernization expertise has also opened broader opportunities for Ensono.

"I want to touch other parts of the IT estate, because I'm not just a mainframe services provider," Klingbeil said. "I know we're known for mainframes, and sometimes it's hard to shake that label, but mainframe services only account for 67% of my revenue."

The convergence of mainframe and AI

While the race for AI capabilities began in the cloud, many enterprises are reconsidering on-premises infrastructure to control costs and safeguard data security. Organizations aiming to migrate off mainframes may also be overly optimistic about the process.

According to a report released by Gartner in June, more than 70% of mainframe exit projects initiated this year are expected to fail.

"The right mainframe strategy depends heavily on the characteristics and complexity of the organizational environment," Gartner VP Analyst Alessandro Galimberti said in the report. "For many mainframe customers, a more effective use of generative AI is to enable in-place modernization rather than accelerate migration off the platform."

Ensono holds a similar view, and this stance predates the current AI boom.

"Mainframes are well-suited for certain workloads that either cannot be replicated in the cloud or would cost three times as much," Klingbeil said. "When migrating off the platform is the right choice, we do it; when in-place modernization is more appropriate, we do that too."

As AI costs and data security concerns intensify, hardware vendors and distributors expect the hybrid strategy to remain solid. TD Synnex CEO Patrick Zammit told Channel Dive that shifting from AI subscriptions to token-based billing models could be a turning point.

"I would say right now that hardware might become attractive again," Zammit said.

Klingbeil added that AI pricing has become a pain point for Ensono.

"It's extremely confusing," he said. "What exactly is a token, why do prices vary, which one is good, and what am I getting out of it?"

AI is also a potential boon for the MSP industry. Service providers use the technology internally while also helping customers deploy use cases.

Ensono jumped in last year, investing $250 million to establish an AI innovation division. The company developed predictive analytics tools for preventing hardware failures, diagnostic AI assistants to help teams identify and resolve technical issues, and an AI application for IT risk assessment called Change Guardian.

Klingbeil said the company has cut the time teams spend handling IT help desk tickets in half. The key to achieving this outcome lies in the data platform the company spent a year building.

"Without high-quality data, AI just gives you more confidence to make mistakes," Klingbeil said.