IBM lost 25.21% on July 14 to close at $217.07, the steepest single-session move in the company’s history and the worst percentage decline for the stock since 1987. The pain was triggered by an unscheduled preliminary Q2 letter from CEO Arvind Krishna disclosing that revenue would come in around $17.2 billion versus the $17.86 billion FactSet consensus, with operating EPS of $2.93 versus $3.01 expected. Software growth, IBM Q2 warning’s central bruise, slowed to 5% from 11% in Q1, well below the full-year target management had put above 10%. Infrastructure revenue fell 7% as IBM Z hardware demand weakened, and enterprise customers reallocated software and mainframe budgets toward supply-constrained AI infrastructure. Roughly $68 billion of market capitalization was wiped in a single trading day. This piece walks through the preliminary numbers themselves, unpacks the capex reallocation mechanic that broke the model, lays out the bull case for a Krishna-led AI consulting rebound in 2027, and then flags the bear case if software growth structurally stays below the 8% line.
The Read
- Preliminary Q2 revenue $17.2B versus $17.86B consensus, EPS $2.93 versus $3.01, both worse than IBM had guided.
- Software growth cut in half from 11% to 5% because enterprise clients rerouted budgets into AI infrastructure.
- Bull case leans on 2027 AI consulting recovery, bear case triggers if software growth stays below 8% full year.
IBM Preannounces Q2 Miss With Software Growth Halved to 5%
Arvind Krishna’s unscheduled letter dropped hours ahead of the regular Q2 print calendar, an unusual move IBM only makes when the shortfall is large enough to require a market pre-alert. The letter walked through preliminary revenue at $17.2 billion versus the $17.86 billion FactSet consensus, a roughly $660 million miss, and operating EPS of $2.93 versus the $3.01 street estimate, an 8-cent gap. The stock reacted the way it usually does when a mega-cap issues its own pre-warning: violently. IBM closed 25.21% lower at $217.07, the steepest single-day drop in the company’s history and its worst percentage decline since 1987.
Software was the wound. Growth halved from 11% in Q1 2026 to 5% in Q2, well short of the greater-than-10% full-year framing IBM had leaned on for the past two quarters. The bull view had priced software as the durable growth engine after the Red Hat integration matured, so the deceleration cuts against the entire narrative multiple. Infrastructure did not help: revenue fell 7% with IBM Z hardware demand noticeably weaker.
Krishna framed the miss as a supply-driven demand reallocation, not a structural loss of share. The distinction matters for how bulls and bears will price the next two quarters. Bulls read a temporary rotation. Bears read a signal that enterprise IT budgets are being permanently reset around AI infrastructure rather than software layers IBM historically dominated, in line with Nvidia’s own AI capex framing at $3-4 trillion by 2030. Neither reading can be fully falsified before Q3 hard numbers land.
Enterprise Clients Reroute Software Budgets to AI Infrastructure
The mechanic Krishna described is specific. Enterprise clients shifted their quarterly capex spend toward servers, storage and memory purchases, aiming to lock supply-constrained AI infrastructure ahead of expected price increases. Money that was going to renew IBM software subscriptions and refresh IBM Z mainframes was pulled forward and pushed into chips and memory. IBM had anticipated some supply chain impact, but not the magnitude of the capex reprioritization that hit its Q2 book.
The reallocation reads as a coherent extension of what Bernstein and other sell-side desks have flagged all year. Even a fraction of the multi-year AI capex pool has to come from budgets historically parked in enterprise software, professional services and legacy infrastructure. IBM sits at the intersection of all three lines.
The chip beneficiaries of the same rerouting have been posting numbers that mirror the IBM print in reverse. Micron jumped 13% earlier this quarter as the AI trade pulled cash from other risk assets, and semi peers with AI memory exposure have been the cleanest beneficiaries of the exact substitution flow Krishna described. On the software side, hyperscaler-owned models are pulling enterprise deals toward the largest AI shops, effectively bidding for the enterprise software wallet IBM has counted on.
One thing to notice is the timing of the pre-alert. IBM chose to disclose before the print date rather than let a full Q2 miss surprise the tape, a tell that management believed the software growth deceleration was material enough to require an early framing. That framing may cushion Q2 print day itself, but it locks Krishna into delivering a credible Q3 recovery signal on the actual earnings call.
Also on CFinance:
- Payout Ratio Reveals What Dividend Yield Hides
- Coinbase Stock at $158, 64% Below Its 52-Week High
- IPO vs SPAC: How Two Paths to Public Markets Diverge
Krishna’s Software Pivot Could Turn 2027 Into an AI Consulting Rebound
The bull case rests on IBM Consulting picking up the slack that IBM Software just posted. IBM Consulting has been positioning through 2026 as an integrator for enterprises that need to graft AI workloads onto existing IBM Z, Db2 and Red Hat estates. If the same clients that rerouted Q2 software renewals into AI infrastructure spend now come back for the integration work in Q4 and 2027, the revenue mix shifts but the top line eventually recovers.
The multiple compression that hit on July 14 also creates a valuation cushion. IBM entered the week trading close to 21x forward earnings, and after the 25% move now trades closer to 16x on the same forward estimates. For bulls, that is a level where a normal software growth reacceleration to 8-9%, halfway back to the 11% Q1 print, would rerate the stock materially without requiring management to defend the 10%+ target it just missed.
The other side of this is that Krishna kept the strategic framing on infrastructure investing rather than defensive cost cuts. If the AI capex boom continues into 2027, IBM’s story becomes “the enterprise partner that survived the software air pocket and now sells the AI integration layer on top of the chips.” That is a credible, if narrow, thesis for the long fundamental holders who use IBM as a diversifier away from the Magnificent 7 concentration risk.
If Software Growth Stays Below 8% Full Year, the Multiple Compresses Further
The bear scenario does not require anything dramatic. It only requires the Q2 software deceleration to hold. If full-year software growth prints between 5% and 7%, IBM structurally loses its “durable software compounder” label and gets rerated as a legacy IT services vendor whose largest hardware line just posted a 7% revenue decline. In that world the 16x forward multiple compresses toward the 12-13x range where the pre-cloud IBM used to trade, and every enterprise deal that Microsoft captures with its 6,000-engineer Frontier program at $25B compounds the shift away from the IBM software wallet.
The bear case also has to price the possibility that enterprise consumption of IBM Z is entering a real terminal-decline phase, not a one-quarter dip. Infrastructure revenue at minus 7% is not a signal one shrugs off, particularly when the pre-alert cited weaker IBM Z as the main driver. Every quarter that IBM Z declines faster than mainframe modernization revenue can offset is a leg down for the segment.
Here is why the tape may punish a soft Q3 more than usual. After a preliminary Q2 warning of this magnitude, sell-side estimates for Q3 will already be revised down. A second miss into those revised estimates would signal that Krishna’s supply-driven demand reallocation framing was too generous, and that a share loss to hyperscalers is the real story. That is the scenario where IBM trades toward the 12x floor and holds there until a credible software reacceleration lands.
More to come.
About the Author
