IBM CEO Arvind Krishna is playing defense. As investors grow anxious about AI eroding the company's legacy software business, Krishna insisted Thursday that the unit will recover within a year, pointing to growing mainframe hardware capacity as a sign of resilience. The comments come as enterprise software giants face mounting pressure to prove their traditional products won't be steamrolled by generative AI alternatives eating into their core businesses.
IBM CEO Arvind Krishna spent Thursday afternoon trying to calm jittery investors who've watched AI anxiety chip away at the company's software business. His message? Don't panic - the recovery's already in motion.
Krishna pointed to a surprising bright spot: mainframe hardware capacity is actually growing, he told investors, suggesting that IBM's core enterprise customers aren't abandoning ship despite the AI revolution swirling around them. The software unit, which has stumbled in recent quarters, should "catch back up" within the next year, Krishna predicted during what sources described as a reassurance campaign.
But the very need for reassurance tells you everything about the moment legacy enterprise software vendors are living through. IBM isn't alone in facing existential questions about whether generative AI will render decades-old software platforms obsolete. The difference is how aggressively Krishna is pushing back against that narrative.
The timing is revealing. Enterprise software companies have spent the past year watching AI-native startups raise billions while their own customers openly question whether to renew expensive licenses or bet on newer AI-powered alternatives. IBM's software portfolio - spanning automation, data management, and security tools - represents a massive revenue stream built on long-term enterprise relationships. Krishna's basically arguing those relationships, and the mainframe infrastructure underneath them, remain rock-solid.
Here's what makes his argument interesting: mainframe capacity growth actually does suggest something. These aren't impulse purchases - enterprises don't expand mainframe infrastructure on a whim. If customers are investing in more capacity, they're planning for continued reliance on IBM's ecosystem, not plotting an exit strategy.
Still, the one-year timeline Krishna floated raises as many questions as it answers. What exactly does "catch back up" mean? Return to previous growth rates? Stop declining? The lack of specifics suggests IBM might still be figuring out how AI reshapes its software business model.
The broader context matters here. While Microsoft and Google have pivoted aggressively into AI-powered productivity tools, IBM's positioned differently. Its enterprise customers move slower, value stability over innovation, and often run mission-critical systems where "good enough" beats "cutting edge." Krishna's essentially betting that conservatism wins - that CIOs at banks and insurance companies won't rip out proven IBM software for shiny AI alternatives.
That's not a crazy bet. But it's also not the growth story investors have rewarded in the AI era. Nvidia didn't become a $3 trillion company by telling shareholders to wait a year for things to stabilize. The enterprise software playbook that made IBM an icon - deep customer relationships, high switching costs, incremental innovation - looks increasingly dated when startups are rebuilding entire categories with AI-first architectures.
Krishna's also threading a needle here. He needs to project confidence that IBM's software business isn't structurally impaired while simultaneously acknowledging enough reality to maintain credibility. Too bullish and he looks out of touch. Too cautious and the stock tanks further.
The mainframe angle is particularly fascinating because it's simultaneously IBM's greatest strength and its biggest albatross. Those systems process unfathomable amounts of critical transactions - credit card payments, airline reservations, insurance claims. They're not going anywhere soon. But "not going anywhere" is different from "growing," and IBM's challenge is converting mainframe stickiness into software expansion rather than managed decline.
What Krishna didn't say might matter more than what he did. No mention of AI revenue targets for the software unit. No specifics on how IBM's embedding AI into existing products versus building new ones. No competitive positioning against Microsoft's Copilot or Google's Workspace AI features. Just a promise that time will heal what ails the business.
Investors will now spend the next year watching whether Krishna's timeline holds. If software stabilizes and mainframe capacity keeps growing, he looks prescient. If the slide continues, that one-year promise becomes the timeline for much harder questions about IBM's software strategy in an AI-dominated landscape.
Krishna's one-year recovery pledge sets a clear clock for IBM's software turnaround, but the real test is whether legacy enterprise vendors can adapt fast enough to stay relevant in an AI-first world. Growing mainframe capacity offers genuine evidence of customer commitment, yet the defensive tone reveals how much pressure traditional software businesses face from AI disruption. IBM's bet on stability and long-term relationships might prove right - or it might just buy time before harder strategic choices become unavoidable. Either way, the next four quarters will show whether IBM's software business can coexist with AI or gets reshaped by it.
