The Reacceleration Thesis Cracks: After a Record Crash and a Cut Guide, We Step Back to Hold on IBM

Results vs. Consensus

Because IBM pre-announced its headline figures on July 14, the meaningful comparison is against the Street's original, pre-warning expectations, which is what the pre-announcement missed and what drove the crash. Against the post-warning revised consensus, the July 22 print was roughly in line.

Year-over-Year Comparison

First-Half 2026 (for context)

Assessment

Revenue. One percent total growth is the weakest print in our coverage, and the composition is worse than the headline. Reported software grew 5%, but management was explicit that organic growth was flat, with the acquired subscription assets (Confluent, HashiCorp) carrying the segment. Transaction Processing fell 9% and IBM Z fell 42%, so the two lines most levered to the mainframe cycle both went sharply negative in the same quarter. The offsetting bright spots, Data up 18% cc and Distributed Infrastructure up 37%, are real, but they were not enough to keep the aggregate above a crawl.

Margins and cash. Here IBM was resilient. Productivity actions expanded operating pre-tax margin 30 bps and adjusted EBITDA margin 20 bps despite the revenue shortfall and continued Confluent dilution, and management raised the full-year margin guide to +100 bps. Free cash flow of $2.5B was down $0.3B year over year, and H1 free cash flow was flat at $4.8B, but management held the full-year guide of roughly +$1B, arguing the first-half drag from inventory buy-aheads, cash taxes, and interest is largely behind it. That is a bigger back-half ask than a quarter ago, but the productivity engine is clearly doing the work it is supposed to do in a down quarter.

EPS. The 5% operating EPS growth flatters the quarter. It is a function of margin and a lower tax rate, not revenue, and it slightly missed the pre-warning consensus. The more honest read is that IBM converted a genuine top-line disappointment into an in-line-ish earnings number, which is a credit to the operating model but does not change the growth question the miss raised.

Segment Performance

Key portfolio KPIs

Software (+5% reported, flat organic): the crown jewel stalls for a quarter

Software is where the thesis lives, and it is where the quarter broke. Reported growth of 5% was carried by acquisitions; organic growth was flat. Management framed the portfolio as roughly 80% high-value recurring revenue (Red Hat subscriptions, HashiCorp, Confluent, watsonx, consumption-based data and automation) and 20% transactional (perpetual licenses and enterprise license agreements concentrated in Transaction Processing and coupled to the mainframe). The recurring 80% grew about 8% and accelerated; the transactional 20% fell high-single-digits as the ELA deals slipped. Underneath the segment, Data grew 18% cc (Confluent's first full post-close quarter), Automation grew 3% cc, Red Hat accelerated a point to +11% cc, and Transaction Processing fell 9% cc.

"About 80% of our annual software revenue is recurring... This revenue grew nicely in the quarter, and our ARR was $24.6 billion, up 8% since last year. The other 20% of software revenue is transactional in nature and is predominantly related to the incumbency with mainframe... Transactional revenue was down high single digits in the quarter, given the shift in buying patterns we have discussed." — James Kavanaugh, CFO

Assessment: The 80/20 framing is analytically useful and largely credible: the durable, subscription part of the software franchise did not deteriorate, and Red Hat's acceleration is a genuine positive. But the full-year Software guide was cut to 6-to-8% from "10-plus percent" one quarter ago, and organic growth of zero is a hard number to explain away as pure timing. Even granting the deferral thesis, the segment that was supposed to be the load-bearing driver of 2026 just delivered its weakest organic quarter in years.

Infrastructure (-7%): the mainframe air-pocket, masked by a record Distributed quarter

Infrastructure fell 7%, with IBM Z down 42% against the +70% z17 launch-quarter comp of a year ago. Management stressed that the cycle remains the strongest on record at roughly 130% program-to-program versus z16, worth more than $1B of incremental hardware and a $3B-plus software stack over the cycle, and that there is no evidence of clients migrating off the platform (85% of installed MIPS are stable or growing, with a 2-to-15x total-cost-of-ownership advantage). The offset was Distributed Infrastructure, which grew 37% for its best quarter on record on Power11 and storage demand, exiting with a record backlog of roughly $500M. On the strength of that, the full-year Infrastructure guide was actually raised to low-single-digit growth from a prior "down low-single-digits."

"Infrastructure revenue declined 7% this quarter. While Z performance was below our expectations through the first 5 quarters of z17 availability, revenue is well ahead of the prior cycles at nearly 130% program-to-program... we see no evidence of clients moving off mainframe." — James Kavanaugh, CFO

Assessment: The Z number looks alarming but is partly a comp-and-timing artifact of the launch-quarter base and the slipped ELAs; the program-to-program metric argues the cycle is intact. The genuinely good news is Distributed Infrastructure, which is emerging as a real AI-adjacent growth vector and let management raise the segment guide even as the mainframe undershot. This is the one place the quarter got structurally better rather than worse.

Consulting (+1% cc): signings grow again, revenue still won't follow

Consulting revenue grew 1% cc, its familiar crawl, but signings grew 6% for a second consecutive quarter, with generative AI now roughly half of signings and more than 30% of backlog. Segment profit margin expanded a healthy 160 bps on productivity. Management held the full-year guide of low-to-mid single-digit revenue growth, arguing backlog quality and GenAI momentum support an acceleration in the back half.

"In Consulting, signings grew 6%, marking our second consecutive quarter of growth... Generative AI represented about 50% of our signings in the quarter and now makes up over 30% of our backlog." — James Kavanaugh, CFO

Assessment: The signings-to-revenue gap is now four quarters old, and patience is wearing. The GenAI backlog mix is encouraging and argues against the AI-cannibalization bear case, but until signings convert to faster revenue, Consulting stays a low-growth ballast rather than a contributor. The margin expansion is the redeeming feature this quarter.

Financing (+12%): small, steady, investment-grade

Financing grew 12% reported to $186M, with a receivables book that is roughly 80% investment grade. It is immaterial to the thesis but a reminder that the captive-finance arm continues to support hardware and software placements at low credit risk.

Assessment: No read-through to the debate. It does its job quietly.

Key Topics & Management Commentary

Overall Management Tone: Confident on strategy, contrite on execution. Having pre-announced, management did not relitigate the numbers; it spent the call defending the "deferred, not destroyed" interpretation and pointing to the parts of the portfolio still growing double digits. The posture was a clear step down from the assured, offensive tone of the April call, where management pressed the AI-tailwind case; here the CEO conceded plainly that "it comes down to execution... that is where we fell short," while insisting conviction in the long-term model is unchanged. The most reassuring notes were on cash and margin, which management defended in quantitative detail; the least convincing was the confidence that a back-half software acceleration to "approaching double digits" is de-risked.

What actually happened in June: a CapEx reallocation, not a demand collapse

The central topic was the mechanism of the miss. In the final weeks of June, a severe shortage in servers, storage, and memory drove hardware prices up sharply (management cited alternate-infrastructure purchases rising roughly 30% in dollar value quarter to quarter), and clients redirected budgets to secure that supply-constrained hardware ahead of further increases. Because large mainframe purchases and their attached software are bought through capitalized enterprise license agreements, tens of those deals slipped, taking Transaction Processing and some Data revenue with them. Management estimated roughly $400-500M of the total miss sat in software.

"Many clients redirected spending towards servers, storage and memory purchases to secure supply-constrained infrastructure ahead of expected price increases. As a result, tens of large deals failed to close on the time lines we expected, accounting for the majority of the shortfall." — James Kavanaugh, CFO

Assessment: The explanation is coherent and externally corroborated by the memory-price environment, and it is a materially better story than share loss or a broad enterprise-demand recession. But it also means IBM's most profitable revenue is hostage to the AI-hardware capital cycle in a way the "AI is a pure tailwind" framing of April did not anticipate. The near-term read-through is that AI CapEx can crowd out IBM's software spend, not just pull it.

Deferred or destroyed: the one-third-already-closed data point

The question the entire investment case now hinges on is whether the slipped deals come back. Management's strongest evidence: roughly a third of the deferred deals had already closed in the first three weeks of Q3, versus a normal expectation of two-thirds to three-quarters closing over six months. The CEO argued clients deferred rather than cancelled because the underlying workloads are mission-critical and deferring only accrues technical debt on a platform they intend to keep.

"About 1/3 of those have already closed. To give you a perspective, normally, we would not expect all of them to close, but we would expect maybe 2/3 to 3/4 of them to close over the next 6 months. So the fact that 1/3 have already closed in the first 3 weeks... gives us a good indication that this was deferral and not destruction." — Arvind Krishna, CEO

Assessment: This is the most important number in the report and it genuinely supports the deferral thesis. It is not yet proof: the pace of conversion over the full quarter, and whether the remainder closes without price concessions, is the swing factor for the back half. It is enough to keep us from an Underperform, not enough to keep us at Outperform.

The guidance cut, unpacked

Management cut full-year revenue growth to 4-to-5% cc from "more than 5%" and Software to 6-to-8% from "10-plus percent." The CFO characterized the low end of each range as a deliberate "anchor" that assumes recent buying behavior persists, holding profit and cash steady, while the high end assumes the slipped deals close and a normal pipeline-conversion resumes (recurring software approaching 10% in H2, Transaction Processing still down low-to-mid-single-digits, treated as a 2027 growth vector). Offsetting the software cut, Infrastructure was raised roughly two points to low-single-digit growth on Distributed strength, and the margin and free-cash-flow guides were held or raised.

"We now expect revenue growth for 2026 in the range of 4% to 5%... We are revising our software growth expectation to a range of 6% to 8% for the full year... we now expect to deliver 100 basis points of operating pretax margin expansion with productivity more than offsetting the revenue-related headwinds." — James Kavanaugh, CFO

Assessment: The structure of the cut is reassuring on cash and unnerving on credibility. Holding free cash flow and raising the margin guide while cutting revenue shows the productivity flywheel absorbing the shock exactly as designed. But raising the Software guide to "10-plus percent" in April and cutting it to "6-to-8%" in July is a large swing in one quarter, and it resets how much benefit of the doubt the next guide earns.

Long-term double-digit software: conviction reaffirmed

Asked directly whether the transactional weakness changes the long-term double-digit software ambition, the CEO was emphatic that it does not, arguing the recurring 80% is already at 8% and approaching 10%, the 80/20 mix keeps shifting toward recurring (it was 60/40 years ago), and inorganic additions (HashiCorp, Confluent, watsonx) grow well above the segment average.

"We and I have complete conviction in the double-digit long-term software growth, let me be clear about that... every single year, that mix of 80-20 used to be 60-40. So as we go over time, that mix is going to climb up and more and more is annuity as opposed to transactional." — Arvind Krishna, CEO

Assessment: The structural argument is sound: a rising recurring mix plus above-average-growth acquisitions is a plausible path back to double digits. The tension is that this path leans increasingly on M&A to carry the number, which sharpens rather than resolves the "how organic is the growth" question the quarter reopened.

Mainframe economics and the Transaction Processing timing question

On how a -42% Z and a -9% TP recover, management leaned on the installed-base and monetization mechanics: 140 million installed MIPS, 85% stable or growing, a 2-to-15x TCO advantage, and the $3-plus of software that follows every hardware dollar over the cycle. The nuance on Transaction Processing is that clients can run on an operating-expense model and defer the capitalized upgrade for years, so the timing of TP monetization is client-controlled, which is why management now treats TP as a 2027 growth vector rather than a 2026 one.

"The more capacity that's in the marketplace, the more monetization opportunity... But when we look at TP and the models that we have done... we've got TP down low single digits to mid-single digits overall because we're going to look at that as a 2027 growth vector opportunity for us." — James Kavanaugh, CFO

Assessment: Pushing the TP monetization into 2027 is prudent and honest, but it also concedes that a chunk of the mainframe software tail that underpinned the bull case is later than previously implied. The installed-base durability is real; the timing of its monetization is looser than the April narrative suggested.

New optionality: Lightwell and quantum

Two forward investments got airtime. Lightwell is a new offering that uses AI to remediate and validate open-source software vulnerabilities, sold at $1M per year, with early marquee adopters concentrated in large banks and a TAM management sizes in the "multiple billions"; more than 7,500 package versions were made available in the first two weeks. Separately, management reiterated a commitment of more than $10B to quantum over five years, including the Anderon foundry with the U.S. Department of Commerce and a roadmap to a large-scale fault-tolerant machine by 2029.

"We believe that it's in the multiple billions of dollars, whether it turns out to be that the TAM... is $5 billion or $10 billion or $15 billion, it's in that range... we're going to go after really fast and hard, leveraging expertise in AI and open source both." — Arvind Krishna, CEO

Assessment: Both are credible long-dated call options that fit IBM's security-and-trust positioning, and Lightwell in particular is a clever monetization of the Red Hat open-source franchise. Neither moves the 2026 numbers, and the $10B quantum commitment is a real multi-year cash claim to note against the maintained free-cash-flow guide.

Productivity and the go-to-market reset

Management is responding to the miss with two levers: accelerated productivity (AI-driven software development, sales and marketing efficiency, supply-chain optimization) that is running ahead of plan and funding the raised margin guide, and a go-to-market expansion beyond the Fortune 1000 into thousands of additional mid-market clients, including forward-deployed engineers aimed at the momentum areas (Red Hat, HashiCorp, Confluent, watsonx, Storage).

"We are also accelerating changes to our go-to-market model. The goal is to expand coverage across thousands of additional clients where our portfolio is highly relevant and where we see significant wallet share opportunities." — Arvind Krishna, CEO

Assessment: The productivity response is exactly right and is why margins and cash held. The go-to-market expansion is a sensible medium-term growth initiative but also an implicit admission that the enterprise-concentrated model left coverage gaps; it will take several quarters to judge, and it carries execution risk of its own after a quarter defined by execution stumbles.

Guidance & Outlook

The shape of the guide is a barbell: the growth lines came down (revenue and software), the profitability and cash lines held or improved, and Infrastructure moved up on Distributed strength. Management was candid that the low end of the revenue range is an "anchor" assuming June's buying behavior persists, while the high end assumes the slipped deals close and pipeline conversion normalizes, with recurring software approaching 10% in the back half. Currency is now expected to be roughly neutral for the full year, with a 1.5-point headwind specifically in Q3.

Implied back-half ramp: Getting from 1% revenue growth in Q2 to a 4-to-5% full year requires the second half to run at roughly mid-single digits, a meaningful acceleration that leans on both a return to ~130% mainframe program-to-program performance and recurring software approaching double digits. The free-cash-flow guide similarly implies a back-half-loaded conversion, with management arguing the first-half inventory, cash-tax, and interest headwinds reverse.

Street at: Consensus had already reset toward the pre-announced figures; the debate now is where in the 4-to-5% range the year lands, which is a direct function of the deal-conversion pace. The maintained free-cash-flow and dividend guides are the anchors the bull case leans on.

Guidance style: Historically conservative, but this quarter tests that reputation in a new way. In April the reiteration-not-raise pattern was a reason to trust the cushion; a cut of this size, a quarter after a raise, is the opposite signal, and it means the back-half acceleration embedded in the high end now has to be earned rather than assumed.

Analyst Q&A Highlights

Deferred versus destroyed, and the milestones to watch

The opening exchange went straight to the question that defines the stock: is the June shortfall a timing issue or a structural shift, and what should investors watch to tell the difference? Management pointed to the one-third-already-closed data point as its confidence signal and framed revenue re-acceleration and mainframe program-to-program performance as the KPIs to track.

Q: "The biggest debate everyone has since the pre-announcement has been, is this demand deferred or destroyed for IBM? So I'd love to understand what gives you the confidence that this is more a timing issue versus a structural shift in enterprise spending? And what milestones should we all be watching for over the next couple of quarters?"
— Amit Daryanani, Evercore ISI

A: "The majority of what didn't happen in the second quarter was large CapEx deals at large clients... about 1/3 of those have already closed... the fact that 1/3 have already closed in the first 3 weeks gives us an indication, not yet full evidence, but a good indication that this was deferral and not destruction." — Arvind Krishna, CEO

Assessment: Management's honesty is notable, calling the evidence "not yet full evidence." The one-third figure is the single best reason to believe the deferral story, and the named milestones (revenue re-acceleration, ~130% mainframe program-to-program) are the right things to hold them to over the next two prints.

Squaring the guidance cut with the deals that already closed

A pointed line of questioning pressed on the math: if some slipped deals have already closed, why cut the guide by roughly a point at the low end, and what does the range actually assume? Management walked through the anchor-versus-acceleration construction of both the revenue and software ranges.

Q: "Your prior guide was 5-plus percent. You're guiding now 4% to 5%... Arvind said that some of those deals have closed. Can you just kind of square up the guide to what this means?"
— Brent Thill, Jefferies

A: "On the low end, it's about 1 point plus takedown. On the high end, give or take, it's pretty damn close to where we were at... The lower end was an anchor. That says nothing changes... The high end says that we basically are approaching double digits in the second half." — James Kavanaugh, CFO

Assessment: The anchor framing is a reasonable way to bound a genuinely uncertain back half, and holding the high end near the prior guide is a subtle vote of confidence. But it also means the reassuring end of the range depends on the same deal-conversion assumption that has yet to fully play out, so the guide is only as good as the deferral thesis.

The path back for Transaction Processing and the mainframe

A recurring worry in Q&A was how the two worst lines, TP at -9% and IBM Z at -42%, recover, and whether the long-term double-digit software ambition still holds. Management defended the mainframe cycle's record program-to-program performance and reaffirmed the long-term software target while conceding TP is now a 2027 story.

Q: "There's 2 lines that are really bothering folks in the guidance. It's TPP and then the mainframe and how you're going to improve from negative 9% and negative 42%... are you still looking for double-digit growth in software long term?"
— Benjamin Reitzes, Melius Research

A: "We and I have complete conviction in the double-digit long-term software growth, let me be clear about that... [on the mainframe] we're still at nearly 130%, a record across the history of IBM of a program to program... That is $1 billion more of mainframe revenue. That is $3-plus billion more of software stack on top of that." — Arvind Krishna, CEO / James Kavanaugh, CFO

Assessment: The program-to-program metric is the strongest defense of the mainframe cycle, and the installed-base economics are real. The concession that TP is a 2027 growth vector is candid but pushes a piece of the software tail further out than the April narrative implied, which is exactly the kind of quiet timing slippage that erodes a growth multiple.

What changed in the final weeks to un-stick the deals

One line of questioning probed the mechanics directly: what happened in the last three to four weeks that let IBM recapture slipped deals so quickly, and whether the company had to discount to do it. Management attributed it to clients recognizing the deferred purchases were mission-critical and that IBM had held price.

Q: "What exactly changed in the last 3 to 4 weeks that catalyzed pushing some of these large transactions that did slip to a successful finish line? And were there any internal or IBM specific efforts... you maybe took a more assertive stance with to recapture these transactions?"
— Fatima Boolani, Citigroup

A: "We have actually held our price so far in the month of July. We have not taken increases... I actually believe that the biggest part is they deferred it and then they realized this is mission critical. All I'm doing is taking on tech debt. I better actually act on this." — Arvind Krishna, CEO

Assessment: That the recaptured deals closed without price concessions is an important, underappreciated detail: it argues the deferral was budget-timing, not a demand or pricing-power problem. It is one of the more genuinely bullish disclosures on the call, and it partly offsets the credibility hit from the cut.

Why organic Data underperformed the rest of recurring software

A sharper question noted that, backing out Confluent's contribution, organic Data revenue was actually down mid-single-digits, an outlier versus the rest of the non-transactional portfolio, and asked why IBM's data software got deprioritized this early in the AI build-out. Management acknowledged Data caught collateral damage from the CapEx-deal slippage and budget prioritization toward Confluent.

Q: "If you take your really bullish comments on Confluent accelerating, you get to organic data revenue growth that was down year-over-year, maybe mid-single digits in the second quarter... why IBM's data software of all things maybe got deprioritized this specific quarter?"
— Erik Woodring, Morgan Stanley

A: "In aggregate, about $400 million to $500 million of the miss was in software overall... data takes what I would call maybe the run by shooting... if Confluent looks more attractive than other things, they are going to buy more of what is the thing that gives them the most near-term value, and we are fine with that." — Arvind Krishna, CEO

Assessment: The candid admission that organic Data was down, even as the reported segment grew 18% on Confluent, is the cleanest illustration of the quarter's core problem: the acquired assets are masking soft organic trends. "We are fine with that" is a defensible near-term posture on incumbency, but it underscores how much of the growth story now rides on M&A rather than the base portfolio.

Sizing the Lightwell open-source security opportunity

The closing exchange turned to strategy and optionality, asking management to size the new Lightwell open-source security offering and explain its importance to IBM's position as the largest enterprise provider of open-source software. The CEO framed it as an AI-enabled remediation "factory" addressing a multibillion-dollar TAM.

Q: "When we're looking at Project Lightwell, could you just talk about... how big of an opportunity this is?... how important solutions in this area in general are for you as the largest enterprise provider of open source software?"
— Matthew Swanson, RBC Capital Markets

A: "We built kind of a factory using AI models to be able to patch open source even when it is not one that we have human expertise in... we believe that this is a multiple billion dollar opportunity, which we're going to go after really fast and hard." — Arvind Krishna, CEO

Assessment: A credible, well-differentiated long-dated option that leverages the Red Hat franchise and IBM's security positioning. It is immaterial to 2026 and will be judged on the client-count milestones management set, but it is exactly the kind of adjacent monetization the platform should be generating.

What They're NOT Saying

1. A full-quarter conversion figure: Management gave the encouraging "one-third closed in three weeks" data point but declined to quantify how much of the slipped pipeline it expects to convert by the end of Q3, or the dollar value still at risk. That is the single number that would most de-risk the back-half guide, and it was left qualitative.
2. Whether the memory/hardware CapEx crowd-out persists: The miss was attributed to a specific June dynamic (clients diverting budget to supply-constrained hardware ahead of price hikes), but management did not address whether that dynamic continues into the second half as memory prices stay elevated, which is the macro risk to the "deferral" framing.
3. The organic software growth rate, stated plainly: Management confirmed organic software was "flat" and organic Data was down, but did not put a clean organic-growth number on the full segment or say when organic re-accelerates, leaving the acquisition-masking question only partly answered.
4. A quantified free-cash-flow bridge for the back half: The full-year free-cash-flow guide was held despite a flat first half, but the specific back-half drivers (inventory flush, cash-tax reversal, working capital) were described directionally rather than sized, making the roughly +$1B target harder to underwrite.
5. The full cost of the go-to-market and talent build: Management announced a coverage expansion to thousands of new clients plus forward-deployed engineers and accelerated productivity, but did not net the investment cost against the productivity savings, so the margin math behind the raised +100 bps guide is taken partly on faith.
6. Any 2027 framing for software: Having reaffirmed the long-term double-digit ambition while cutting 2026 to 6-to-8% and pushing TP monetization into 2027, management offered no bridge for how and when software re-rates to double digits, which is the crux of whether this is a one-quarter air-pocket or a slower structural trajectory.

Market Reaction

• The damage was done on July 14, not the print: IBM pre-announced the shortfall in a pre-market CEO letter on July 14 and fell 25.2% that session, from $290.23 to $217.07 (-$73.16), on 67.4M shares (6.5x the 30-day average). It was the largest single-day decline in the company's recorded history and erased roughly $67B of market value.
• Pre-print setup: The stock then leaked a further ~5% into the formal print, closing July 22 at $205.77, a fresh 52-week closing low. Entering the print it was down 30.5% year-to-date and 27.0% over twelve months, against an S&P 500 up 9.5% year-to-date, with a 52-week closing range of $205.77 to $329.23.
• Reaction to the formal print (July 23): Shares closed at $206.65, up 0.4% (+$0.88), on 15.3M shares (1.3x average). The muted move reflects that the July 22 release only confirmed the pre-announced numbers and formalized the guide cut, while pairing it with a maintained free-cash-flow and dividend guide and a raised Infrastructure outlook.

The two-stage reaction is the whole tell. The market repriced IBM violently on the surprise itself, then treated the confirmatory print as a non-event bordering on relief. That is consistent with a stock that has moved from a growth-reacceleration narrative to a show-me-the-conversion narrative: the sellers who needed to leave left on July 14, and what remains is a value-and-cash-flow holder base waiting on evidence. The stability on the print does not vindicate the quarter; it simply reflects that the bad news was already fully in the price. Whether $206 is a floor or a way-station depends entirely on the Q3 deal-conversion print.

Street Perspective

Debate: Is the June shortfall deferred or destroyed demand?

Bull view: The bull case, and management's, is that this is a budget-timing air-pocket: clients diverted CapEx to secure scarce hardware, the deferred deals are mission-critical, a third already closed within three weeks without price concessions, and the recurring 80% of software kept growing. On this read the back half snaps back and the crash created a rare entry point.

Bear view: The bear camp argues that "deferred" is what every company says after a miss, that a 42% mainframe decline and flat organic software are too large to be pure timing, and that if AI-hardware CapEx is structurally crowding out enterprise software budgets, the pressure recurs rather than reverses.

Our take: The evidence tilts toward deferral, the one-third-closed-without-discount data point is the best proof, but it is not yet conclusive, and the burden is now on management to demonstrate conversion in Q3. We are not willing to underwrite Outperform on an unproven snap-back a quarter after a large guide raise became a guide cut. This is precisely the kind of resolution that a Hold, not a bet, is designed to wait for.

Debate: Does the productivity-and-cash resilience make IBM a buy on the dip?

Bull view: The bulls note that IBM defended earnings (operating EPS +5%), raised the margin guide to +100 bps, held free cash flow at roughly +$1B, and maintained the dividend, all in its worst revenue quarter in our coverage; at roughly 8% free-cash-flow yield and a 3.3% dividend near a multi-year low, the downside is well anchored and the cash story is intact.

Bear view: The bears counter that cash and dividend safety is necessary but not sufficient, that IBM has spent a decade looking cheap on cash flow while the growth engine disappointed, and that a stock re-rating lower on a broken growth story is the textbook value-trap pattern, not a dip to buy.

Our take: Both are right, which is why this is a Hold and not an Underperform or an Outperform. The cash-and-dividend floor is real and prevents a bearish call at this valuation; the broken growth catalyst and dented credibility prevent a bullish one. The free-cash-flow yield gets you paid to wait, but there is no reason to pay up before the growth question resolves.

Debate: How much of IBM's growth is genuinely organic?

Bull view: The bulls point out that Red Hat accelerated to +11%, HashiCorp set a bookings record, the recurring software base grew 8% and is heading toward 10%, and the rising subscription mix structurally lifts the organic rate over time even if a single quarter looks acquisition-heavy.

Bear view: The bears seize on management's own admission that organic software was flat and organic Data was down mid-single-digits, arguing the reported 5% software growth is an M&A mirage and that IBM increasingly needs deals to hit its numbers, which is neither cheap nor sustainable at scale.

Our take: The quarter genuinely reopened this debate that April had seemed to close. The recurring base is healthier than the organic-flat headline implies, but the reliance on Confluent and HashiCorp to carry reported software growth is now unmistakable, and it raises the bar for the organic portfolio to prove itself. Until organic software re-accelerates, the growth-quality discount the market is applying is defensible.

Model Update Needed

Valuation impact: At the July 23 close of $206.65, IBM trades at roughly 17x our ~$12.0 FY26 operating EPS estimate, an ~8% free-cash-flow yield on the maintained ~$15.7B guide, and a 3.3% dividend yield. On a 16-18x multiple that reflects the growth reset against the offsetting free-cash-flow quality, fair value sits around $190-215, essentially in line with the current price. That is a materially cheaper stock than the ~18x we upgraded at in April, but it is no longer the clear bargain the depressed price makes it look: the free-cash-flow yield anchors the downside near current levels while the reset growth trajectory and the modest P/E leave little room to pay up until conversion evidence arrives. The setup is balanced, which is a Hold.

Thesis Scorecard Post-Earnings

Overall: Thesis weakened. The growth pillars that justified our April upgrade (software reacceleration, the mainframe multiplier, AI as an unambiguous tailwind) all moved to Challenged, and a new execution-credibility risk emerged. The durability pillars (productivity, cash, dividend, Red Hat, valuation floor) held, which is what keeps this a Hold rather than a sell. The net is a business whose downside is well protected but whose near-term growth catalyst has been interrupted and whose management just spent a quarter of hard-won credibility.

Action: Downgrade to Hold from Outperform. This reverses the April upgrade, which was explicitly a bet on software reaccelerating with a raised guide, a bet the quarter contradicted. We stay constructive on the cash flow, the dividend, and the deferral-not-destruction interpretation, but we will not pay up for it before Q3 shows the slipped deals converting and organic software re-firming. We would move back to Outperform on evidence of that conversion plus a return to double-digit recurring software; we would move to Underperform if the deferral proves to be destruction (deals do not close), recurring software decelerates, or the free-cash-flow and margin guides slip. Until then, the roughly 8% free-cash-flow yield gets holders paid to wait.