IBM slashed its full-year revenue growth forecast to 4–5 percent after enterprise customers redirected data center budgets away from mainframe purchases and toward servers, storage, and memory needed for AI workloads. The 42 percent quarterly contraction in mainframe sales is not a cyclical pause—it is a permanent reallocation of customer spending, and the numbers prove it.
The picture is budget cannibalization, not budget contraction. Customers are not spending less on data center infrastructure; they are spending differently. IBM’s distributed infrastructure business—Power servers and storage—grew 37 percent year-over-year to an all-time record for that division. The same customers who deferred mainframe upgrades were simultaneously accelerating AI-compatible hardware purchases. Net effect: a 7 percent decline in overall infrastructure revenue to $3.8 billion, dragged down entirely by the mainframe line. IBM’s problem is upstream of the software-replacement concern affecting companies like Salesforce, Workday, and Snowflake: AI infrastructure spending is absorbing the capital budgets that would otherwise fund traditional enterprise hardware, including the highest-margin mainframe upgrades. Same customers; wallet share is shifting and not shifting back.
The sequencing dynamics behind the numbers confirm a deliberate reallocation under AI-driven urgency. The divergence between mainframe sales at minus 42 percent and distributed infrastructure at plus 37 percent is too large to be explained by normal product-cycle variation. Chief Executive Arvind Krishna called the environment “early innings of a structural shift for business”—an understatement. The mainframe contraction is the leading edge of a permanent reallocation of enterprise computing spending toward infrastructure designed for AI workloads. Customers who have experienced the priority queuing and budget flexibility that AI hardware procurement demands will not return to the cadence of traditional mainframe upgrade cycles. AI hardware commands higher priority in purchasing queues because supply constraints have created expectations of near-term price increases, and as Chief Financial Officer James Kavanaugh told the Wall Street Journal, those constraints are not expected to ease: “Everyone in the industry is saying this will extend for a period of time.”
Kavanaugh also disclosed that IBM failed to close “tens of deals” in the second quarter—customers who might have purchased mainframe systems instead redirected those budgets to AI infrastructure. The company has since recovered roughly a third of those deals, but the displacement has runway.
IBM earned $2.2 billion on revenue of $17.2 billion in the quarter and still expects full-year free cash flow to increase by approximately $1 billion from last year—reflecting a cost structure and software portfolio absorbing the mainframe decline without immediate financial distress. Shares rose as much as 3.5 percent in after-hours trading following the formal earnings release, but that recovery sits against the July 14 preliminary disclosure that erased approximately $67 billion in market value and sent shares down 25 percent in one of the worst single-day declines in IBM’s history.
The July market reaction created boardroom pressure that extends beyond a single quarter. The company, which employs 260,000 people, has become the subject of breakup and activist speculation. The board of directors is expected to meet in late July and faces a central question: can Krishna’s multi-year investment thesis—AI and hybrid cloud—deliver fast enough to satisfy investors who absorbed a 25 percent single-day decline? The H2 mainframe recovery narrative depends on the assumption that customers who deferred for AI hardware will return once supply constraints normalize. The quarter’s data undermines that assumption. Customers are adapting procurement workflows to prioritize AI infrastructure; the behavioral shift outlasts any particular supply bottleneck. The H2 recovery narrative is a hope, not a forecast grounded in what the quarter actually showed. Kavanaugh maintains that mainframes run over 70 percent of the world’s transaction volumes by value, arguing, “Mainframe has always been the most secure, resilient, scalable system in the world.” That may be true, but the data says customers are buying something else.
Watch items: Q3 mainframe deal closure rates and whether they return to pre-displacement levels; whether distributed infrastructure segment growth decelerates as AI hardware supply constraints ease, confirming the current surge is displacement-driven; the board’s late-July meeting output and whether it accelerates AI commercialization, considers portfolio restructuring to shed legacy drag, or attempts to ride out the displacement without structural change; and the point at which “structural shift” becomes IBM’s own language for a permanent mainframe contraction, reshaping the company’s valuation multiples.
This analysis draws on publicly reported earnings data and statements made by IBM executives; it does not allege intent on the part of any named individual or organization.
Analytical techniques used in this piece
This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.
- Relationship Mapping
- Extracts the network of ties among people, institutions, and entities.
- Scenario Planning
- Builds a small set of distinct, plausible futures to plan against.
- Systems Dynamics (Structural)
- Maps a system’s structure — stocks, flows, and the architecture that shapes its behavior.