IBM delivered a sobering update. The tech giant slashed its full-year revenue growth forecast to between 4% and 5%. This marks a step down from its earlier projection of more than 5% expansion. The culprit? A stunning 42% plunge in mainframe sales during the second quarter.
Customers redirected cash. They funneled it toward AI infrastructure instead. Servers, storage and memory chips claimed priority as prices climbed and supplies tightened. The shift caught IBM off guard in late June. It wiped out gains the company had posted since rolling out its latest Z systems hardware a year earlier.
Mainframe Demand Collides With AI Hardware Frenzy
The numbers tell a stark story. Total revenue for the quarter ended June 30 rose just 1% to about $17.2 billion. That figure landed roughly $660 million below Wall Street expectations. Adjusted earnings per share came in near $2.93. The infrastructure segment, home to the mainframe business, dropped 7% to $3.8 billion.
Transaction-processing software tied to those mainframes suffered too. It stayed essentially flat. CFO Jim Kavanaugh pointed directly at the redirection of capital expenditures. “In the last few weeks of June, we saw clients shift their quarterly capex spend toward servers, storage, and memory purchases to secure supply-constrained infrastructure ahead of expected price increases,” he explained in comments reported by Reuters.
But here’s the twist. This weakness follows periods of strength. In late 2025, IBM Z mainframe revenue had surged. The z17 launch drove gains as banks and insurers embraced its AI capabilities for secure, on-premises processing. Now that cycle appears to be winding down faster than anticipated. And AI itself is accelerating the change. Enterprises face ballooning costs for GPUs and specialized chips. Some saw data-center expenses jump 15% to 30%. So they paused mainframe refreshes. They deferred software expansions. One analyst called it a classic case of AI eating the budget before the metal even arrives.
Shares reacted violently at first. When IBM issued preliminary results on July 14, the stock plunged more than 25% in a single session. That drop ranked as the company’s worst in decades. By the time full earnings landed on July 22, much of the pain had been priced in. Shares rose about 3% in extended trading. Investors seemed relieved the actual miss matched the warning.
Yet the episode raises bigger questions. IBM has poured tens of billions into software. Acquisitions of Red Hat, HashiCorp and others aimed to reposition the company as a hybrid-cloud and AI leader. Those bets produced solid growth elsewhere. Software revenue climbed 5% in the quarter. Consulting advanced nearly 3%. But the high-margin transaction software linked to mainframes dragged the segment’s overall performance. Kavanaugh trimmed the full-year software growth outlook to 6% to 8%. He insisted the cut tied solely to infrastructure softness. The rest of the business performed well, he added.
CEO Arvind Krishna struck a similar tone. He described the mainframe weakness as temporary. Refresh cycles should resume once AI build-outs normalize and supply chains stabilize. “Clients reprioritized their spend to the hardware,” Krishna told CNBC. The company still expects to generate an extra $1 billion in free cash flow this year. Cost cuts, tighter supply-chain management and lower third-party spending will help. Headcount will stay roughly flat.
Even so, the Starbucks example lingers. The coffee chain reportedly eyes replacing certain IBM software with in-house AI tools. That application costs Starbucks about $2 million annually. Kavanaugh conceded it sits in the category most open to disruption. Most of IBM’s portfolio, however, runs closer to the infrastructure layer. Replacing it proves far harder. The company continues to modernize its Z platform. A partnership with Arm seeks to bring contemporary workloads onto mainframes while preserving the security and efficiency that made them indispensable for decades.
Recent coverage reinforces the pattern. NDTV Profit highlighted how the mainframe dip confirms legacy businesses face pressure even as IBM doubles down on AI. The Information noted the revenue target reduction reflects customers choosing AI hardware over traditional systems. On X, analysts and executives echoed the theme. One post observed that “AI isn’t killing it—budgets are shifting.” Another framed it as infrastructure moving faster than oversight.
Longer term, IBM bets its mainframe franchise will adapt. The z17 already integrates specialized AI acceleration. Banks use it to run large language models without shipping sensitive data to the cloud. Demand for such transactional AI remains high. Yet near-term volatility could persist. Memory shortages and GPU prices show little sign of easing quickly. Enterprises must balance immediate AI projects against sustaining core systems that still process the majority of the world’s financial transactions.
Kavanaugh pushed back against broader fears that generative AI will hollow out IBM’s software empire. “Most of IBM’s software sits close to enterprise infrastructure and data, making it far harder to replace than the applications most vulnerable to AI disruption,” he told The Next Web. That distinction matters. It separates IBM from pure-play application vendors now scrambling to defend their turf.
Still, the quarter exposed limits to the company’s transformation narrative. Investors who bid up the stock nearly 30% earlier in the year had bet on consistent software acceleration. The miss, however slight, shattered that confidence. Software came in at $7.39 billion. Analysts had modeled $7.43 billion. Hybrid cloud, including Red Hat, grew 16%. Those figures looked respectable in isolation. They simply couldn’t offset the infrastructure hole.
IBM now accelerates cost-saving moves. It continues to acquire talent and technology in data and AI consulting. The purchase of Hakkoda expands its ability to help clients integrate AI into core operations. Such steps aim to smooth the transition. They cannot erase the reality that hardware cycles and software renewals remain tightly linked in the mainframe world. When one falters, the other feels it immediately.
Wall Street will watch the next few quarters closely. If mainframe demand rebounds as Krishna predicts, the current dip may prove a bump. If AI capex continues to crowd out other spending, IBM may need to rethink its growth algorithm once again. For an enterprise technology leader that once defined the industry, the message is clear. Adaptation never stops. And sometimes the very technologies you champion create the biggest near-term headaches.
One thing seems certain. The mainframe isn’t going away. Its revenue just hit a speed bump on the road to AI relevance. How IBM steers through that curve will test the strength of its hybrid strategy and the depth of its enterprise relationships.
WebProNews is an iEntry Publication