IBM · Q2 2026 · reported 2026-07-22 · revenue $17.16bn
Assessment
Q2 2026 is the quarter IBM missed its own expectations, and the AI disclosure moved with it. No level is given for AI software, where Q1 gave trailing revenue north of ; the CEO says only that the watsonx portfolio is growing much faster than software overall, and the filing no longer names generative AI products among the reasons Data revenue grew. Generative AI consulting is back on a bookings basis, about of signings and over of backlog, with the Q1 revenue run-rate above not repeated and consulting revenue up adjusted for currency. The ledger's sizes, and , carry the Q1 trailing-year figure and the Q1 floor on the share of consulting revenue forward. No channel is sized by management.
Management attributes most of the shortfall to large deals deferred as clients moved capital budgets to servers, storage and memory ahead of shortages and price increases, and puts the software part of the miss at about to ; Transaction Processing revenue fell . Management does not call the shortage an AI effect; analysts did. The ledger reads the shift as a confound on AI software and registers no toll: a channel needs the company or its filing to name AI as the cause. The same shift lifted Distributed Infrastructure , which the CEO credits to AI adoption and data growth and the filing to clients buying ahead. About of the slipped deals had closed by the time of the call.
Lightwell was launched: a subscription at per client per year for open-source packages remediated with AI models, after a new AI model sped up the discovery of vulnerabilities. It became generally available after the quarter ended, so its revenue for the quarter is left unsized, beside a stated build commitment of with no period; the company also says it is investing in forward deployed engineers. On the mainframe an attach rate appears, nearly of z17 customers investing in the AI accelerator, in a quarter when derived IBM Z revenue changed by ; a share of customers by count sizes nothing, so the mainframe reading is unsized this quarter.
On its own costs the dollar figures of Q1 are gone and the line moves instead. The filing credits productivity and transformation actions with points of lower selling, general and administrative expense, , against in Q1, and consulting gross margin rose to from ; the filing's expense bullet does not name AI, its MD&A says the Client Zero savings come in part from 'embedding AI in our own workflows' with no share, and the call lists AI beside third-party spend, sales efficiency and supply chain. AI is one of several levers of the expense benefit and of the consulting margin gain, and nothing separates its part, so neither is sized; the developer saving is sized on its own line. The developer productivity rate of Q1 is not repeated while research and development expense rose ; the filing credits that line to AI, hybrid cloud and quantum together, so research and development spending on AI is unsized in both quarters. Nothing is said about agents and application software, or about what the company pays for models.
Sized channels against the income statement, Q2 2026
7 of 14 channels sized
Each blue mark is one channel's dollars for the quarter; a bar is the range of an estimate. Grey marks are the company's reported lines. The distance between them is the point: how large the AI channel is next to the line it sits in.
Total revenuereported line
$17.16bn
Total costreported line
$7.25bn
Selling, general and administrativereported line
$4.98bn
Research and developmentreported line
$2.31bn
Interest expensereported line
$486mn
Restructuring chargesreported line
$30mn
Consulting engagements management tags as generative AIrevenue in · our inference
AI software: watsonx platform, agents, assistants and orchestrationrevenue in · our inference
Lightwell commitment: frontier AI capabilities and engineers for open-source remediationspend · our inference
Model, token and inference cost of the company's AI products and internal AI usespend · our inference
Software development cost avoided through the company's own AI coding system (Bob)cost displaced · our inference
Red Hat, data and automation software demand attributed to customers' AI adoptionrevenue in · our inference
Forward deployed engineers and specialized talent for clients' AI deploymentsspend · our inference
$100k$1mn$10mn$100mn$1bn$10bn$100bn
AI channel, dollars for the quarter low to high of an estimate reported lineLog scale: each gridline is ten times the one before.
New money and old money, Q2 2026
2 new9 expanded3 relabelled
Each channel is tagged once for whether its money existed before language models, from the company's annual report and call at the start of the period. A bar splits one flow's sized dollars by that tag. The incremental total is the part that would not be there without the models: a new channel counts in full, an expanded one only for what AI added, a relabelled one at zero.
Flow, sized total
Split by novelty
Incremental total
Paid for AI$85mn to $1.61bn sized
new $22mn to $269mnexpanded $63mn to $1.34bn
Incremental total $22mn to $1.61bnpoint $91mn$324mn in 2 channels has no traced baseline
Cost displaced by AI$8.3mn to $146mn sized
expanded $8.3mn to $146mn
Incremental total $8.3mn to $146mnpoint $46mn
Revenue arriving through AI$1.33bn to $2.14bn sized
expanded $255mn to $630mnrelabelled $1.07bn to $1.51bn
Incremental total $0 to $630mnpoint $0$413mn in 1 channel has no traced baseline
The sized total counts every channel the company credits to AI, including relabelled money that existed before language models and the ledger's own estimates for it. The incremental total counts relabelled channels at zero. A flow is split by layer where its dollars sit at more than one: end use, compute sold to builders, and hardware. The same dollar can be a buyer's spend, a cloud's revenue and a chipmaker's revenue, so the layers are never added together.
Paid for AI4 channels · $85mn to $1.61bn sized · $22mn to $1.61bn incremental · 1 not sized
research
Research and development spending on AI
Not sized
The filing again credits the line to investment to drive innovation in AI, hybrid cloud and quantum together (claim c32), and no source gives AI's share; nothing separates AI's part, so the channel is left unsized with no ballpark on a judgment share (the methodology rule for AI named beside another cause, rules sweep of 2026-10-06). The ceiling is the research and development line, . The former estimate, ibm-2026-cq2-f78, was an assumed AI share of that line.
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: described· motive: exploratory· before LLMs: expanded
Research and development expense rose to , attributed as in Q1 to organic and acquired investment in AI, hybrid cloud and quantum (claim c32). No AI share is given. AI is named beside hybrid cloud and quantum and nothing separates its part, so the reading is unsized; the line is its ceiling. The Lightwell engineering effort launched in the quarter sits inside it.
Evidence: 3 quotes, 4 figures, 3 confounds, 1 from before coverage
The part of research and development expense that builds AI: models, the watsonx platform, agents, AI features in existing software and AI hardware. The filing names AI first among the investments behind the line's growth and never gives its share, and acquired engineering staff are in the same line.
Why this motive
The filing again frames the spending as investment to drive innovation in AI, hybrid cloud and quantum (claim c32), and the CFO speaks of investing behind an AI opportunity (claim c34): the research framing.
Before LLMs: expanded
Research, development and engineering expense was in FY2024, up , which the annual report attributed to investments in AI, hybrid cloud and quantum and to the next IBM Z cycle. The share that is AI work is not reported at the anchor or since. The size is the whole of an activity that existed before.
“Research, development and engineering (RD&E) expense increased 10.4 percent in 2024 versus 2023, primarily driven by investments to drive innovation in AI, hybrid cloud and quantum, as well as in Infrastructure ahead of our next IBM Z cycle in 2025.”
Research and development, prior-year quarter · 2025-CQ2
Research and development, year-over-year growth · 2026-CQ2
Research and development, year-over-year increase · 2026-CQ2
Reported line it is matched to
Research and development expense rose to . The filing names AI first among its investment areas (claim c32).
2026-CQ2: 2025-CQ2: 2026-CQ2:
What else could explain it
acquisition: Engineering staff acquired with Confluent, DataStax and HashiCorp are in the line.
line composition: Hybrid cloud, quantum, mainframe and storage development share the line with AI; the company also announced a large quantum investment plan in the quarter.
other: The filing names hybrid cloud and quantum beside AI as what the investment drives (claim c32); no source separates AI's part.
Quotes
“Research and development (R&D) expense increased 10.2 percent and 10.8 percent in the second quarter and the first six months of 2026, respectively, driven by continued organic and inorganic investments to drive innovation in AI, hybrid cloud and quantum.”
“The Software gross profit margin decline for the second quarter and the first six months of 2026 was primarily driven by investments in our portfolio innovation, and product mix.”
Q1 2026described · described, no size · exploratory
Q2 2026described · described, no size · exploratory
vendor bill
Model, token and inference cost of the company's AI products and internal AI use
0.13% to 1.6% of the quarter’s revenue
Incremental total: counts in full.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: described· motive: exploratory· before LLMs: new
Use of models widened with Lightwell, which the CEO describes as a factory using AI models and the letter as backed by frontier AI capabilities (claims c35, c55); no cost is given. Software gross margin fell to from , which the filing attributes to investments in portfolio innovation and product mix (claim c33) without naming AI. The size, , is the ledger's own, built as in Q1. The counterparty is mixed: third-party frontier model providers, and the compute that runs the company's own and open-weight models.
Evidence: 4 quotes, 3 figures, 3 confounds, 1 from before coverage
What the company pays to run AI, for its customers and for itself: third-party frontier models, its own and open-weight models, and the compute behind them. Management describes using several model providers and its own smaller models and says nothing about what that costs; the cost would sit in software cost and in operating expense.
Why this motive
Management describes building with AI models, now including a factory of them for open-source remediation (claim c35), and says nothing about what the models cost or whether the cost is recovered in price. AI named with no measure and no line moving is exploratory; the sources do not contradict each other.
Before LLMs: new
The anchor reports no model, token or inference bill. It describes the company's own small Granite models and cheaper inferencing as a selling point, beside a software gross profit margin of in FY2024.
“We enable cheaper inferencing built for hybrid cloud architectures with our Red Hat AI portfolio. We provide small, open Granite models that deliver better performance at a fraction of the price.”
Software gross profit below what the prior-year margin would have given · 2026-CQ2
Reported line it is matched to
Software gross profit was , below what the prior-year margin of would have given. The filing attributes the decline to investments in portfolio innovation and product mix (claim c33).
2026-CQ2: 2026-CQ2: 2026-CQ2: 2025-CQ2: 2026-CQ2:
What else could explain it
acquisition: Amortization of acquired intangibles and Confluent's hosting cost are in software cost and lower the margin.
mix shift: The fall in high-margin transactional software in the quarter lowers the margin without any AI cost; the filing names product mix.
bundling: The frontier models behind Lightwell sit inside its stated commitment and are not added here.
Quotes
“The Software gross profit margin decline for the second quarter and the first six months of 2026 was primarily driven by investments in our portfolio innovation, and product mix.”
“The portfolio of AI offerings we have built, including cost efficient, fit-for-purpose open-source models deployed in hybrid environments, is focused on helping businesses scale AI and generate return through productivity improvements and automation.”
“Lightwell is a $5 billion commitment backed by new frontier AI capabilities and a global force of more than 20,000 engineers creating a trusted enterprise clearinghouse to address open source software vulnerabilities.”
Lightwell commitment: frontier AI capabilities and engineers for open-source remediation
0.36% to 7.3% of the quarter’s revenue
Incremental total: counts at zero at the point and in full at the high end, with no traced baseline.
our inferencedisclosure: bounded· motive: exploratory· before LLMs: expanded
The letter describes Lightwell as a commitment of , backed by frontier AI capabilities and more than engineers (claim c55), and the filing as an investment combined with a global workforce of engineers (claim c57). The total bounds the spending and comes with no period, and how much is new money rather than existing engineers' time is not said. The size, , takes the quarter's share of the commitment from the fixed undated band and is the ledger's own. It sits inside research and development spending on AI and is not added to it. The counterparty is mixed: the company's own and Red Hat engineers, and outside frontier model capability. The channel this reading was counted inside (ai-research-and-development) is unsized this quarter, so the overlap no longer holds and this reading counts in totals on its own evidence (overlap rule, 2026-10-06).
Evidence: 4 quotes, 2 figures, 3 confounds, 2 from before coverage
What the company commits to building and running Lightwell: frontier AI capabilities and a large force of its own and Red Hat engineers. Management states a total commitment with no period, and much of it may be existing engineering time pointed at the new product. It sits inside research and development spending on AI.
Why this motive
The filing calls Lightwell an investment and the product has no revenue yet (claim c57): spending ahead of the market it is meant to open.
Before LLMs: expanded
Engineering spend on open-source software sits in research, development and engineering expense, in FY2024. The anchor shows no clearinghouse for open-source vulnerabilities and no commitment of this kind. The size is the whole of an activity that existed before.
“Research, development and engineering (RD&E) expense increased 10.4 percent in 2024 versus 2023, primarily driven by investments to drive innovation in AI, hybrid cloud and quantum, as well as in Infrastructure ahead of our next IBM Z cycle in 2025.”
“Cybersecurity risk to the company and its customers also depends on factors such as the actions, practices and investments of customers, contractors, business partners, vendors, the open source community and other third parties, including, for example, providing and implementing patches to address vulnerabilities.”
relabel: A force of that many engineers is larger than any new hiring could supply in a quarter; much of the commitment may be existing Red Hat and IBM engineering time counted under a new name.
line composition: No reported line isolates the program; engineering sits in research and development and model usage in cost.
other: The commitment has no stated period, and the program ran for only part of the quarter.
Quotes
“What we did was 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.”
“The same tools that are being used to attack all code, both proprietary and open source, can also be used to effectively be those experts and build those out.”
“Lightwell is a $5 billion commitment backed by new frontier AI capabilities and a global force of more than 20,000 engineers creating a trusted enterprise clearinghouse to address open source software vulnerabilities.”
“In the second quarter, we launched Lightwell, an investment to establish a trusted enterprise clearinghouse, combined with a global workforce of engineers, to address open-source software vulnerabilities at scale, which became generally available in July.”
Forward deployed engineers and specialized talent for clients' AI deployments
0% to 0.52% of the quarter’s revenue
Incremental total: counts at zero at the point and in full at the high end, with no traced baseline.
our inferencedisclosure: described· motive: exploratory· before LLMs: expanded
The call and the release say the company is investing in more specialized technical and client-facing talent, including forward deployed engineers, as clients move AI to enterprise-scale deployment (claims c60, c61). No count, cost or line is given. The size, , is the ledger's own, from assumed headcount and cost.
Evidence: 2 quotes, 2 confounds, 1 from before coverage
What the company spends on specialized technical and client-facing staff, including forward deployed engineers, to move clients' AI from experiments to deployment. Management describes the investment with no count or cost.
Why this motive
The spending is described as an investment made as AI adoption moves from experimentation to deployment (claim c60), ahead of the revenue it is meant to bring.
Before LLMs: expanded
Client-facing technical staff existed at the anchor: the FY2024 annual report describes adding consulting and technical skills during the year, inside selling, general and administrative expense of in Q1 2024. No forward deployed engineering group is named. The size is the whole of an activity that existed before.
“In 2024, we focused on adding skills in key areas such as consulting and technical expertise, while also scaling our capacity in strategically important markets.”
line composition: The staff would sit in selling, general and administrative expense or in consulting cost, neither of which isolates them.
other: The same paragraph describes a wider change to sales coverage that is not specific to AI.
Quotes
“As AI adoption moves from experimentation to enterprise-scale deployment, we are also investing in more specialized technical and client-facing talent, including forward-deployed engineers.”
“As AI adoption moves from experimentation to enterprise-scale deployment, the company is also investing in more specialized technical and client-facing talent, including Forward Deployed Engineers.”
Cost displaced by AI3 channels · $8.3mn to $146mn sized · $8.3mn to $146mn incremental · 2 not sized
back office · cheap to verify
Operating cost avoided through the company's own AI-enabled transformation (Client Zero)
Not sized
The selling, general and administrative benefit is credited to productivity and transformation actions in which AI is one lever beside third-party spend, procurement, sales efficiency, supply chain and workforce actions (claims c38, c43 and c45), and no share is given for AI's part (the methodology rule for AI named beside another cause).
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: direction only· motive: exploratory· before LLMs: expanded
No dollar figure for productivity is given this quarter; the Q1 figures, since 2023 and more in 2026, are not repeated. Management says the actions are ahead of plan and lists them: AI and automation at greater scale, lower third-party spend, sales and marketing efficiency, AI in software development, supply chain and services delivery (claims c38, c39, c40). The filing measures the benefit at points of selling, general and administrative expense, on the prior-year line, against in Q1, without naming AI in that bullet (claim c43); its MD&A says the company is 'driving efficiency and cost savings with our Client Zero approach, leveraging technology and embedding AI in our own workflows' and gives no share (claim c45). AI is one of several levers and nothing separates its part, so the channel is left unsized; the software development lever is sized on its own line.
Evidence: 13 quotes, 11 figures, 5 confounds, 3 from before coverage
Cost the company says it avoids by rebuilding its own operations with data, automation and AI, which management calls its productivity engine and its Client Zero approach. Management states the program's savings as a cumulative figure and an annual addition and names AI as one of several levers beside third-party spend, sales and marketing, supply chain, tax and procurement work. The filing's expense bullet reports the benefit in points of selling, general and administrative expense without naming AI, while its MD&A says the Client Zero savings come in part from embedding AI in the company's own workflows, with no share. Nothing separates AI's part of that benefit, so the channel is left unsized under the rule for AI named beside another cause (2026-10-06). Workforce rebalancing and spend cuts are confounds and are never counted as an AI saving.
Why this motive
The displaced line shrinks: selling, general and administrative expense other than charges and amortization fell to from , and the filing credits productivity actions with points (claim c43). Productivity is being accelerated while revenue growth slowed, which could read narrative-defensive; management does not attribute workforce actions to AI and the line movement is in the filing, so efficiency is kept. The result is credited to several causes together, so it does not meet the efficiency tell for any one of them (the methodology's motive table): exploratory.
Before LLMs: expanded
Productivity initiatives predate the coverage window: the FY2024 annual report describes scaling AI within IBM, automation and AI-driven efficiencies, and an HR assistant resolving of low-level inquiries, alongside workforce rebalancing charges of in the year. Selling, general and administrative expense was in Q1 2024. The size is the change AI made, not the whole line.
“We remain focused on our productivity initiatives as we digitally transform our business processes and scale AI within IBM. This includes simplifying our application and infrastructure environments, aligning our teams by workflow and enabling a higher value-add workforce through automation and AI-driven efficiencies.”
“For example, we have resolved 94 percent of low-level HR inquiries with our AskHR assistant, built on watsonx, freeing up HR professionals to focus on more complex issues.”
“IBM is increasingly applying AI-based technologies, including generative AI, to its services and products, to how it delivers offerings to IBM clients, and to its own internal operations.”
Increase in selling, general and administrative expense the filing attributes to acquired businesses, in points · 2026-CQ2
Productivity savings since 2023, cumulative, as stated · as-of 2026-03-31
Additional productivity savings expected in 2026 · FY2026
Reported line it is matched to
Selling, general and administrative expense fell to from ; the filing sets a productivity benefit of points, , against increases from acquired businesses and amortization (claims c43, c44).
transformation program: The actions are a cost program in a quarter that missed on revenue: third-party spend, sales and marketing and supply chain sit beside AI. Workforce rebalancing charges were against and are not attributed to AI.
acquisition: Acquired businesses added points to the line (claim c44).
fx: The filing credits currency beside productivity for the fall in operating expense (claim c46).
other: Lower revenue-linked spending in a quarter of missed deals would also lower the line.
other: Spend cuts, procurement, sales and marketing efficiency, supply chain and workforce rebalancing are named beside AI as levers of the same benefit, and nothing separates AI's part.
Quotes
“At the same time, we are accelerating productivity actions across the company, spanning both spend reduction initiatives and actions designed to drive growth. These include leveraging AI to improve software development productivity, increasing the effectiveness of our sales and marketing organization, and accelerating our supply chain.”
“These actions include deploying AI and automation at greater scale across the company, reducing third-party spend, improving sales and marketing efficiency, using AI to drive more efficient software development, optimizing our supply chain, and enhancing services delivery.”
“While revenue dynamics are creating margin pressure for the year, the pace of our productivity actions have exceeded our expectations. As a result, we now expect to deliver 100 basis points of operating pre-tax margin expansion, with productivity more than offsetting the revenue-related headwinds.”
“Despite this shortfall, productivity actions were ahead of plan, providing the flexibility to absorb Confluent-related dilution, continued investing for growth, and expanded adjusted EBITDA and operating pre-tax margins by 20 and 30 basis points, respectively.”
“Productivity Enables Investment and Value: IBM is accelerating productivity by scaling software development leveraging AI, increasing the effectiveness of its sales and marketing organization, and optimizing its supply chain.”
“In addition, we are taking action to accelerate our revenue growth and profitability, driving productivity across the company with AI and automation, and heavily investing in commercializing innovation at speed and scale.”
“Higher operating expenses from acquired businesses, as a result of our continued investment to drive our hybrid cloud and AI strategy (4 points); and”
“AI is a powerful productivity driver for our clients and for IBM. We are transforming our enterprise operations, driving efficiency and cost savings with our Client Zero approach, leveraging technology and embedding AI in our own workflows. Our developer workforce is using IBM Bob, our AI-based software development system that automates the full software lifecycle, driving developer productivity and predictable enterprise costs.”
“Total operating (non-GAAP) expense and other (income) decreased 1.5 percent year to year, driven by savings from productivity actions and the effects of currency, partially offset by our investments in portfolio innovations.”
“Although our second-quarter 2026 results fell short of our expectations, we continue to focus on the fundamentals of our business by enhancing our go-to-market model, advancing our high-growth portfolio, and accelerating our innovation and productivity actions which span both spend reduction initiatives and actions designed to drive growth.”
Q1 2026quantified · described, no size · exploratory
Q2 2026direction only · described, no size · exploratory
engineering · cheap to verify
Software development cost avoided through the company's own AI coding system (Bob)
0.05% to 0.85% of the quarter’s revenue
Incremental total: counts in full.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: described· motive: narrative-defensive· before LLMs: expanded
The call and the release name AI in software development as a productivity action and give no rate (claims c37, c38, c41); the gain of stated in Q1 is not repeated, and the filing says only that the developer workforce uses Bob (claim c45). The size, , carries the Q1 rate and is the ledger's own. It is one lever of the company-wide program; its base, research and development expense, lies outside the selling, general and administrative benefit the program channel is sized on, so the two are added.
Evidence: 4 quotes, 3 figures, 3 confounds, 3 from before coverage
Engineering cost the company avoids because its developers use Bob, its own AI software development system. Management gives a productivity rate and no dollar figure, and the research and development line it would show in is rising. It is one lever of the company-wide productivity program, sized on research and development expense, apart from the program's selling, general and administrative benefit.
Why this motive
AI in software development is named among the productivity actions (claims c37, c38) with no rate this quarter, while research and development expense rose : a stated lever with no displaced line shrinking, so the less durable motive is kept.
Before LLMs: expanded
Software development is a cost the anchor shows inside research, development and engineering expense, in FY2024 and in Q1 2024. The anchor names code assistants as products for clients and says AI is being applied to internal operations, with no measure for the company's own developers. The size is the change AI made, not the whole line.
“Research, development and engineering (RD&E) expense increased 10.4 percent in 2024 versus 2023, primarily driven by investments to drive innovation in AI, hybrid cloud and quantum, as well as in Infrastructure ahead of our next IBM Z cycle in 2025.”
“We remain focused on our productivity initiatives as we digitally transform our business processes and scale AI within IBM. This includes simplifying our application and infrastructure environments, aligning our teams by workflow and enabling a higher value-add workforce through automation and AI-driven efficiencies.”
“IBM is increasingly applying AI-based technologies, including generative AI, to its services and products, to how it delivers offerings to IBM clients, and to its own internal operations.”
Research and development, year-over-year growth · 2026-CQ2
Average productivity gain of the company's developers using Bob, as stated · 2026-CQ1
Reported line it is matched to
Research and development expense rose to : the line again moves against the claim, which the filing explains by investment (claim c45).
2026-CQ2: 2025-CQ2: 2026-CQ2:
What else could explain it
line composition: Research and development expense holds hardware, research and acquired staff beside the developers who use Bob.
acquisition: Acquired engineering teams raise the line and hide any saving in it.
other: The rate rests on a single statement from the prior quarter, with no method given.
Quotes
“At the same time, we are accelerating productivity actions across the company, spanning both spend reduction initiatives and actions designed to drive growth. These include leveraging AI to improve software development productivity, increasing the effectiveness of our sales and marketing organization, and accelerating our supply chain.”
“These actions include deploying AI and automation at greater scale across the company, reducing third-party spend, improving sales and marketing efficiency, using AI to drive more efficient software development, optimizing our supply chain, and enhancing services delivery.”
“Productivity Enables Investment and Value: IBM is accelerating productivity by scaling software development leveraging AI, increasing the effectiveness of its sales and marketing organization, and optimizing its supply chain.”
“AI is a powerful productivity driver for our clients and for IBM. We are transforming our enterprise operations, driving efficiency and cost savings with our Client Zero approach, leveraging technology and embedding AI in our own workflows. Our developer workforce is using IBM Bob, our AI-based software development system that automates the full software lifecycle, driving developer productivity and predictable enterprise costs.”
Consulting delivery cost avoided through AI (IBM Consulting Advantage)
Not sized
The consulting margin gain is credited to productivity actions generally, and AI is named as one lever of them (claims c38 and c49); nothing separates AI's part (the methodology rule for AI named beside another cause).
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: direction only· motive: exploratory· before LLMs: expanded
Consulting margins moved this quarter: gross margin to from , worth at the prior-year margin, and segment margin to (claim c48). The CFO lists enhancing services delivery among the AI and automation actions (claim c38); the filing credits productivity actions and does not name AI as the cause (claim c49). A direction and a rate are given for the margin, none for AI's part. The margin gain is credited to productivity actions, of which AI is one lever, and nothing separates AI's part, so the channel is left unsized.
Evidence: 4 quotes, 5 figures, 3 confounds, 3 from before coverage
Delivery labor the consulting business avoids by using AI agents and reusable assets on its Consulting Advantage platform. It would show in the consulting gross margin, which the filing explains by productivity actions without naming AI. Part of the work is code, which is cheap to check; process and advisory work is less so. It is one lever of the company-wide productivity program, read on consulting cost of revenue apart from the program's selling, general and administrative benefit, and unsized while nothing separates AI's part.
Why this motive
Consulting gross margin rose to from and segment margin to from , which management credits to productivity actions (claims c48, c49). The result is credited to several causes together, so it does not meet the efficiency tell for any one of them (the methodology's motive table): exploratory.
Before LLMs: expanded
Consulting delivery cost is what lies between consulting revenue of and a gross profit margin of in FY2024. The anchor already describes IBM Consulting Advantage as an AI delivery platform changing how consultants work, and credits margin to productivity actions without naming AI. The size is the change AI made, not the whole line.
“We remain focused on our productivity initiatives as we digitally transform our business processes and scale AI within IBM. This includes simplifying our application and infrastructure environments, aligning our teams by workflow and enabling a higher value-add workforce through automation and AI-driven efficiencies.”
“IBM is increasingly applying AI-based technologies, including generative AI, to its services and products, to how it delivers offerings to IBM clients, and to its own internal operations.”
“IBM Consulting brings speed and scale to innovative solutions that combine industry, domain, and hybrid cloud knowledge together with AI-powered assets, such as IBM Consulting Advantage, a first of its kind AI delivery platform designed to deliver solutions at scale and realize faster time to value, transforming how our consultants work.”
Consulting cost (revenue less gross profit) · 2026-CQ2
Reported line it is matched to
Consulting gross profit was on revenue of , above what the prior-year margin would have given. The filing attributes the margin to productivity actions, partly offset by investments in innovation (claim c49).
2026-CQ2: 2026-CQ2: 2026-CQ2: 2025-CQ2: 2026-CQ2:
What else could explain it
transformation program: The filing credits productivity actions generally; workforce rebalancing and utilization are part of them and are not attributed to AI.
mix shift: A shift toward higher-margin engagements raises gross margin without any change in delivery cost.
fx: Currency lowered reported consulting revenue growth and moves margin through geographic mix.
Quotes
“These actions include deploying AI and automation at greater scale across the company, reducing third-party spend, improving sales and marketing efficiency, using AI to drive more efficient software development, optimizing our supply chain, and enhancing services delivery.”
“Consulting gross profit, segment profit and respective margin performance in the second quarter and first six months of 2026 primarily reflect the benefits of the productivity actions we have taken, partially offset by investments in innovation.”
“In Consulting, AI is both a growth driver and a productivity engine. Our experts are helping clients design and execute AI strategies by leveraging the IBM Consulting Advantage platform, an AI delivery platform designed to implement solutions at scale, transforming how our consultants work. As agents take on more work, delivery becomes faster, more software driven, and more scalable.”
Q1 2026described · described, no size · exploratory
Q2 2026direction only · described, no size · exploratory
Revenue arriving through AI6 channels · $1.33bn to $2.14bn sized · $0 to $630mn incremental · 3 not sized
product revenue
AI software: watsonx platform, agents, assistants and orchestration
1.5% to 3.7% of the quarter’s revenue
Incremental total: counts at zero at the point and in full at the high end, with no traced baseline.
our inferencedisclosure: direction only· motive: offensive· before LLMs: expanded
No level is given for AI software this quarter. The CEO says the watsonx portfolio is growing at much higher rates than software overall (claim c2), software grew adjusted for currency with organic growth flat (claim c9), and the filing no longer lists generative AI products among the reasons Data revenue grew (claim c6). The cumulative book of business is still defined in the release's second exhibit and no figure for it is given (claim c5). The size, , carries the trailing-year figure stated in Q1 through the fixed annual-plan band, with the ledger's own growth since. The CFO now places the generative AI portfolio in the recurring part of software revenue (claim c3), where the Q1 arithmetic and the book's definition pointed to the transactional part.
Evidence: 16 quotes, 7 figures, 4 confounds, 4 from before coverage
Software revenue from the products management groups as its AI platform, agents, assistants and orchestration: the watsonx platform, watsonx Orchestrate, the assistants, and the Bob development system. Through fiscal 2025 management measured it inside a cumulative generative AI book of business that adds transactional software revenue since inception to new contract value and consulting signings; from Q1 2026 it gives a trailing-year revenue figure instead. Neither is a quarter's revenue, so the quarter is always the ledger's estimate.
Why this motive
Separately priced AI products that the CEO says are growing at much higher rates than software overall (claim c2): movement management attributes to AI, with no figure this quarter. The Q1 measure, trailing revenue north of growing north of , is not repeated.
Before LLMs: expanded
At the anchor the company already sold generative AI software, watsonx, Concert and its AI assistants, inside a software segment of in FY2024, and its assistant for customer care sat in a Data and AI category that grew . The only level given was a cumulative book of business above , weighted towards consulting. No line reports the AI products, so no earlier quarter of them can be traced. The size is the whole of an activity that existed before.
“Our investments in generative AI are contributing to growth, as we had strong demand for our generative AI products such as watsonx, Concert and our AI assistants.”
“Data & AI revenue increased 1.6 percent as reported (2.2 percent adjusted for currency), with strong growth in Data Fabric and our AI assistant for Customer Care, driven by client demand for our watsonx platform offerings, and strength in asset and supply chain management software which helps clients run sustainable operations.”
“Inception to date, our book of business related to watsonx and Generative AI is greater than $1 billion, with sequential quarter-over-quarter growth. Similar to last quarter, this remains weighted towards consulting.”
“So AI book of business, I think you nailed it in your question. It's one, on a consulting perspective, it's our signings book of business overall, and on our software, it's our subscription, our SaaS, and perpetual licenses.”
Software revenue, year-over-year growth adjusted for currency · 2026-CQ2
Estimated AI software revenue as a share of software revenue · 2026-CQ2
AI platform, agents, assistants and orchestration software revenue, trailing twelve months, floor · TTM as-of 2026-03-31
Software revenue shortfall against the company's expectations, low end of the stated range (the unit, millions of dollars, is implied) · 2026-CQ2
Software revenue shortfall against the company's expectations, high end of the stated range (the unit, millions of dollars, is implied) · 2026-CQ2
Share of the slipped large deals closed in the first three weeks of the next quarter, approximate · as-of 2026-07-22
What else could explain it
relabel: A different basis in each period: a cumulative book through 2025, trailing-year revenue in Q1, and a growth comparison with no level in Q2. None is restated on another's basis.
line composition: The products sit in the Data and Automation lines, where a full quarter of Confluent now explains the growth the filing reports.
other: Transactional software revenue fell in the quarter as large deals slipped; how much AI software was inside those deals is not said.
other: Clients redirected capital budgets to servers, storage and memory ahead of shortages and price increases, and deals deferred as a result account for most of the shortfall (claims c62, c66, c29); the CEO puts the software part at about to (claim c63), and about of the slipped deals closed early in the next quarter (claim c65). Management does not attribute the shift to AI; analysts on the call did. The ledger records it here and registers no toll, since a channel needs the company or its filing to name AI as the cause.
Quotes
“Watsonx Orchestrate is the control plane that helps clients build, manage, and govern agents, which, combined with Red Hat, gives clients a foundation to run inference and applications on any infrastructure. Bob is our entry point into the developer ecosystem, helping clients build enterprise-ready AI applications and agents while creating a natural pathway to adoption of watsonx Orchestrate and IBM's broader AI platform.”
“As an example, we brought Hashi in and we accelerated it. We brought Confluent in and we accelerated it. Our organic watsonx portfolio is also growing at much higher rates than the aggregate.”
“80% high-value recurring revenue, and that is, think of our portfolio, Red Hat, subscriptions, our acquisitions that are mostly subscription consumption-based, our GenAI portfolio and watsonx, and parts of our data automation portfolio that operate on consumption models.”
“High-Growth Portfolio: Areas of IBM's software business that help clients manage, deploy and build AI-ready solutions, like Red Hat, the watsonx portfolio, HashiCorp, and Confluent continue to deliver strong performance.”
“It is calculated as inception to date Software transactional revenue, plus new SaaS Annual Contract Value and Consulting signings related to specific offerings. Since second-quarter 2023, approximately one-fifth of this book of business comes from Software, and the remaining four-fifths is Consulting.”
“Data revenue increased 18.9 percent as reported (18.4 percent adjusted for currency) reflecting the contribution from recent acquisitions, primarily Confluent.”
“In Software, IBM watsonx provides a robust portfolio of AI products for developing AI apps, managing data, and governing the entire lifecycle of AI models and AI agents, allowing clients to move from pilots to production with full control over cost, security, sovereignty, and performance.”
“Value will increasingly shift towards the orchestration and data layers so that clients can optimize outcomes, cost, and governance across multiple models and agents and keep control of their proprietary data.”
“We have held the view for a while that the unprecedented investment in AI infrastructure and models will increase pressure on enterprises to generate meaningful returns from that spend.”
“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 timelines we expected, accounting for the majority of the shortfall.”
“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. This dynamic impacted client buying patterns. While we anticipated some supply chain related impact in our expectations, we did not anticipate the magnitude of the capex reprioritization. In addition, clients were distracted with rapidly-evolving, industry-wide cybersecurity concerns in the quarter.”
“The fact that a third have already closed in the first three weeks gives us a indication, not yet full evidence, but a good indication, that this was deferral and not destruction.”
“I think that our clients themselves had not really thought through that some of the alternate purchase they were doing were increasing 30% in dollar value quarter-to-quarter. When they were faced with that issue, then they decided to move budget to those areas where they were having that extreme rise.”
Q2 2026direction only · our inference · offensive · $255mn to $630mn
product revenue
Consulting engagements management tags as generative AI
6.2% to 8.4% of the quarter’s revenue
Incremental total: counts at zero.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: quantified· motive: narrative-defensive· before LLMs: relabelled
The quarter's measures are back on a bookings basis: generative AI was about of consulting signings and is over of backlog (claim c10), which on reported figures is of signings and at least of backlog. The revenue run-rate given in Q1, above , is not repeated, and the filing attributes consulting revenue to application modernization, data transformation and cybersecurity demand (claim c14). The state stays quantified because the shares come with reported bases; the stated level is of bookings, not revenue. The size, , carries the Q1 floor of over of consulting revenue as its low end and is the ledger's own.
Evidence: 8 quotes, 9 figures, 3 confounds, 7 from before coverage
Consulting revenue from engagements management counts as generative AI: designing and scaling AI for clients, and delivery that uses generative AI inside client contracts. Management has measured it as signings inside the cumulative book of business, as a share of backlog and signings, and once as an annualized revenue rate. The consulting line it sits in has grown at about the same low rate throughout, so the tag is read as existing consulting work under a new label.
Why this motive
AI talk heavy relative to any quantified dollar: generative AI is about of signings, up from about , and over of backlog (claim c10), and no revenue figure is given, on a consulting line that grew adjusted for currency. The shares are bookings; the offensive tell needs revenue that moves, so narrative-defensive is taken.
Before LLMs: relabelled
Consulting revenue was in FY2024 against in FY2023, growth of adjusted for currency, with data and technology transformation projects focused on AI and analytics already credited for 2023. On the Q1 2024 call the CFO confirmed an analyst's estimate that generative AI was below of a consulting backlog of about , and in his next sentence put it at mid- to high-single digits.
“Inception to date, our book of business related to watsonx and Generative AI is greater than $1 billion, with sequential quarter-over-quarter growth. Similar to last quarter, this remains weighted towards consulting.”
“So AI book of business, I think you nailed it in your question. It's one, on a consulting perspective, it's our signings book of business overall, and on our software, it's our subscription, our SaaS, and perpetual licenses.”
“During 2024, we operated in a dynamic macroeconomic environment following our strong performance in 2023, as clients reprioritized their IT spend toward digital transformation and AI initiatives for cost optimization and operational efficiency.”
“At the same time, we saw both a lengthening of backlog duration, driven by large-scale digital transformations, and a reduced level of revenue realization in the quarter as clients tighten discretionary spending.”
Generative AI share of consulting signings, approximate · 2026-CQ2
Generative AI share of consulting backlog, floor · as-of 2026-06-30
Generative AI consulting signings in the quarter, the stated share applied to reported signings (bookings, not revenue) · 2026-CQ2
Generative AI consulting backlog, the stated share applied to reported backlog, floor · as-of 2026-06-30
Consulting signings · 2026-CQ2
Consulting signings, year-over-year growth adjusted for currency · 2026-CQ2
Consulting backlog at quarter end · as-of 2026-06-30
Consulting generative AI revenue, annualized run-rate, floor · as-of 2026-03-31
Generative AI share of consulting revenue, floor · 2026-CQ1
Reported line it is matched to
Consulting revenue was against , up as reported and adjusted for currency. The filing names application modernization, data transformation and cybersecurity demand (claim c14) and gives no generative AI figure.
2026-CQ2: 2025-CQ2: 2026-CQ2: 2026-CQ2:
What else could explain it
relabel: What is quantified changed from a revenue run-rate in Q1 to shares of signings and backlog in Q2; signings are bookings and are not revenue.
mix shift: Consulting revenue rose adjusted for currency while generative AI reached about of signings: the tagged work appears to replace other consulting work inside a flat line.
other: The estimate rests on a share of revenue stated in the prior quarter; nothing this quarter confirms it.
Quotes
“It is calculated as inception to date Software transactional revenue, plus new SaaS Annual Contract Value and Consulting signings related to specific offerings. Since second-quarter 2023, approximately one-fifth of this book of business comes from Software, and the remaining four-fifths is Consulting.”
“Generative AI represented about 50% of our signings in the quarter and now makes up over 30% of our backlog, underscoring the demand for AI-powered transformations that extend beyond technology modernization into core business operations.”
“In consulting, the quality of our backlog and momentum in GenAI continue to support an acceleration in revenue growth to low to mid-single digits for the year.”
“The revenue performance in Consulting was driven by demand for application modernization, data transformation and cybersecurity services as clients balance the need to increase productivity through AI with the need to strengthen resiliency and manage risk.”
“As clients move from pilots to enterprise-wide deployment, they are increasingly turning to IBM Consulting to re-engineer business processes and unlock productivity and new business value through AI, automation, and digital labor.”
“In Consulting, AI is both a growth driver and a productivity engine. Our experts are helping clients design and execute AI strategies by leveraging the IBM Consulting Advantage platform, an AI delivery platform designed to implement solutions at scale, transforming how our consultants work. As agents take on more work, delivery becomes faster, more software driven, and more scalable.”
Red Hat, data and automation software demand attributed to customers' AI adoption
0.03% to 0.42% of the quarter’s revenue
Incremental total: counts at zero.
Matched line moved : the whole line, not this channel.
our inferencedisclosure: described· motive: exploratory· before LLMs: relabelled
The AI framing of infrastructure software demand continues without a measure (claims c4, c8, c17). The filing attributes Data growth of to recent acquisitions, primarily Confluent, where Q1 also named generative AI products (claim c6), and the CFO attributes Red Hat's growth to subscriptions and consumption services (claim c16). The CEO says the rest of the Data line was hurt by the deferred deals (claim c18). The size, , is the ledger's own, on the same assumed share as in Q1.
Evidence: 8 quotes, 6 figures, 4 confounds, 2 from before coverage
The part of demand for the company's non-AI infrastructure software that management attributes to customers adopting AI: more data consumed from internal systems, more containers and automation as agents multiply applications. It includes the acquired Confluent and DataStax products, which are data companies and not AI companies on this ledger; their revenue is an acquisition confound here and nothing of their cost is credited to AI. Management gives no measure of the AI-attributed part.
Why this motive
The release groups Red Hat, watsonx, HashiCorp and Confluent as software for AI-ready solutions (claim c4) and the CEO says value is shifting to the layers the company supplies (claim c8), with no figure for the AI part. AI named as the cause with no measure and no line moving is exploratory; no narrative-defensive tell of its own is quoted.
Before LLMs: relabelled
Red Hat, automation and data software were on offer at the anchor and growing without an AI explanation: Red Hat revenue rose in FY2024 on application modernization and automation demand, and OpenShift exited the year at of annual recurring revenue.
“Within Hybrid Platform & Solutions, Red Hat revenue increased 11.4 percent as reported (12.0 percent adjusted for currency), which reflects the continued demand for our hybrid cloud solutions as clients are prioritizing application modernization on OpenShift containers and Ansible automation to optimize their IT spending and reduce operational complexity.”
“As generative AI deployment accelerates alongside traditional workloads, developers are working with increasingly heterogeneous, dynamic, and complex infrastructure strategies.”
Data revenue, year-over-year growth as reported · 2026-CQ2
Software revenue, year-over-year growth adjusted for currency · 2026-CQ2
Reported line it is matched to
Hybrid Cloud and Data revenue rose together. The filing attributes Hybrid Cloud growth to subscriptions and consumption services and Data growth to acquisitions, primarily Confluent (claim c6).
2026-CQ2: 2025-CQ2: 2026-CQ2: 2025-CQ2: 2026-CQ2:
What else could explain it
acquisition: A full quarter of Confluent is inside the Data increase, and the filing names it as the primary cause.
line composition: The Data line also holds the AI products sized in the AI software channel.
relabel: Red Hat, data and automation products predate the AI framing of their demand.
other: Deals deferred late in the quarter lowered the organic part of the Data line (claim c18).
Quotes
“As an example, we brought Hashi in and we accelerated it. We brought Confluent in and we accelerated it. Our organic watsonx portfolio is also growing at much higher rates than the aggregate.”
“High-Growth Portfolio: Areas of IBM's software business that help clients manage, deploy and build AI-ready solutions, like Red Hat, the watsonx portfolio, HashiCorp, and Confluent continue to deliver strong performance.”
“Data revenue increased 18.9 percent as reported (18.4 percent adjusted for currency) reflecting the contribution from recent acquisitions, primarily Confluent.”
“Value will increasingly shift towards the orchestration and data layers so that clients can optimize outcomes, cost, and governance across multiple models and agents and keep control of their proprietary data.”
“Red Hat growth accelerated one point sequentially to 11%, driven by improvement in the subscription piece of the business and stable growth in consumption-based services.”
“When we think of those large CapEx deals, while it is majority, but think majority as being 50%-60% TP software, there is a lot of data also in there.”
IBM Z capacity and accelerators bought to run AI on the mainframe
Not sized
The quarter's new measure is a share by count of customers, nearly of z17 customers investing in AI capabilities with the Spyre Accelerator (claim c19), beside a capacity growth multiple for one cohort and capacity growth credited to AI, analytics and Linux workloads together (claim c23). A share of customers by count separates nothing in dollars, so the channel reads directional and gets no ballpark (the methodology's count-only rule, rules sweep of 2026-10-06). The former estimate, ibm-2026-cq2-f76, was an assumed AI share of derived IBM Z revenue, , the ceiling. Q1 stays sized because the CEO named AI alone as a kind of capacity sold; the step to unsized comes from this quarter's wording, not from a change at the company.
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: direction only· motive: offensive· before LLMs: expanded· layer: hardware
The CFO gives an attach rate for the first time: nearly of z17 customers are investing in AI capabilities with the Spyre Accelerator (claim c19), a share of customers by count with no accelerator price or revenue; a second statement in the answers words it differently (claim c22). IBM Z revenue, derived at , changed by on the anniversary of the z17 launch and on deals that slipped (claim c24). Management sees no evidence of clients moving off the mainframe, with clients holding of installed capacity maintaining or growing it (claim c21). The share of customers is quoted and the reading is unsized; the derived line is its ceiling.
Evidence: 7 quotes, 7 figures, 3 confounds, 2 from before coverage
Mainframe hardware and the software priced on its capacity that clients purchase in order to run AI inferencing beside their transactions: the Spyre accelerator on z17, on-chip inferencing, and the capacity growth management associates with watsonx Code Assistant for Z. IBM Z revenue is not reported in dollars; the ledger derives it from the reported Hybrid Infrastructure total and the reported growth rates. The size is the AI-attributed share of that line, which the product cycle mostly drives; it is a level of AI-related mainframe revenue, including AI inferencing demand that predates LLMs, not an increment against the pre-LLM rate.
Why this motive
A disclosed attach rate, nearly of z17 customers investing in the Spyre Accelerator, beside the capacity multiple of for code assistant users (claim c19).
Before LLMs: expanded
At the anchor IBM Z already carried on-chip acceleration for AI inferencing and the company was working with over clients on AI on z16, for fraud and anomaly detection, work that predates LLMs. The annual report credited watsonx Code Assistant for Z as a contributor to Transaction Processing revenue of . Hybrid Infrastructure revenue was in FY2024, with IBM Z changing by late in the z16 cycle. No figure traces the AI part of mainframe revenue at the anchor, so the channel is sized as a level with no baseline. The size is the whole of an activity that existed before.
“IBM Z is uniquely positioned for AI, with the first processor design, with on-chip acceleration for real-time AI inferencing. In fact, we're working with over 100 clients on the application of AI on Z16. Use cases range from fraud detection to anti-money laundering to anomaly detection.”
“The performance in 2024 is the result of the combination of clients' growing capacity demands, solid renewal rates, and increased contribution from our generative AI products, including watsonx code assistant for Z.”
Share of z17 customers investing in AI capabilities with the Spyre Accelerator, approximate ceiling · as-of 2026-06-30
Mainframe capacity growth of clients using watsonx Code Assistant for Z, as a multiple of other clients' growth · 2026-CQ2
z17 revenue against z16 at the same point in the program, approximate · as-of 2026-06-30
Share of installed mainframe capacity held by clients maintaining or growing it · as-of 2026-06-30
IBM Z revenue, year-over-year change as reported (negative: a decline) · 2026-CQ2
IBM Z revenue, derived from the Hybrid Infrastructure totals and the two growth rates · 2026-CQ2
IBM Z revenue, prior-year quarter, derived from the Hybrid Infrastructure totals and the two growth rates · 2025-CQ2
Reported line it is matched to
Hybrid Infrastructure revenue was against , with IBM Z changing by ; the derived IBM Z decline is . The filing attributes it to performance below expectations and a strong prior-year launch (claim c24).
other: The product cycle and the quarter's slipped deals drive the line: IBM Z revenue changed by against the launch quarter, a derived decline of .
line composition: IBM Z revenue is derived, not reported, and accelerators are not reported apart from systems.
other: The attach rate is a share of customers, not of revenue; no accelerator price is given.
Quotes
“We are seeing strong early adoption of our AI innovations, with nearly 50% of z17 customers investing in AI capabilities with Spyre Accelerator, and clients deploying watsonx Code Assistant for Z are growing MIPS capacity three times faster than those who are not.”
“While clients continually evaluate workload placement, we see no evidence of clients moving off the mainframe. z17 remains at nearly 130% program to program, well ahead of z16, which was our strongest on record.”
“We have got clients that have already purchased over 50% of our Spyre Accelerator, and those clients that have purchased that are growing MIPS capacity, the way we monetize value, by over three times faster than others.”
“By the way, that is coming in new AI workloads, analytics workloads, Linux-based workloads, and those MIPS are growing program to date over 15%-20% install capacity.”
“Within Hybrid Infrastructure, IBM Z decreased 42.0 percent as reported (41.8 percent adjusted for currency) reflecting performance below expectations in the current period, and also comparing to a historically strong prior-year z17 launch.”
“Earlier this month, we introduced a smaller LinuxONE system that allows clients to address data center space and cost constraints while offering the security, resiliency, real-time inferencing the Z platform can deliver.”
Q1 2026direction only · our inference · offensive · $8.3mn to $124mn
Q2 2026direction only · described, no size · offensive
product revenue
Storage and Power demand attributed to AI
Not sized
The CEO credits the growth to AI adoption and the growth of enterprise data together (claim c26), and the letter and filing to clients buying ahead of shortages and price increases (claims c28, c29); nothing separates AI's part (the methodology rule for AI named beside another cause).
Matched line moved : the whole line, not this channel.
described, no sizedisclosure: direction only· motive: unknown· before LLMs: relabelled· layer: hardware
Distributed Infrastructure grew , its strongest quarter on record, to a derived , and exited with an order backlog of about (claim c31). The CEO attributes the growth to AI adoption and the growth of enterprise data (claim c26); the letter and the filing attribute it to clients securing servers, storage and memory ahead of shortages and price increases (claims c28, c29). A rate is now given for the line and none for the AI part. The derived increase was ; nothing separates AI's part from data growth and pre-buying, so the channel is left unsized.
Evidence: 6 quotes, 4 figures, 4 confounds, 2 from before coverage
Storage and Power server revenue that management attributes to AI: storage for the data AI needs, and, from Q2 2026, clients buying servers and storage ahead of shortages and price increases. Distributed Infrastructure revenue is not reported in dollars; the ledger derives it from the Hybrid Infrastructure total and the reported growth rates. Management gives no measure of the AI-attributed part.
Why this motive
Separate causes are given for the same growth and cannot be told apart: the CEO names AI adoption and data growth (claim c26), while the letter and the filing name clients buying ahead of shortages and price increases (claims c28, c29). A line growth rate attributed to AI would read offensive; with a second cause stated and no measure of the AI part, the motive stays unknown.
Before LLMs: relabelled
Storage and Power were established lines at the anchor: Distributed Infrastructure grew in Q1 2024 and in FY2024, and both the call and the annual report already credited storage demand to preparing data for generative AI, with no measure of that part.
“Clients are also looking to our storage offerings for data curation, model building, and fine-tuning in support of generative AI.”
“Storage revenue performance was driven by growth in high-end storage tied to the z16 platform and solutions tailored to protect, manage and access data for generative AI.”
Distributed Infrastructure order backlog at quarter end, approximate · as-of 2026-06-30
Reported line it is matched to
Distributed Infrastructure revenue grew , a derived increase of , which the filing attributes to clients redirecting spending to scarce infrastructure and to its storage and Power products (claims c29, c30).
2026-CQ2: 2025-CQ2: 2026-CQ2: 2026-CQ2: 2026-CQ2:
What else could explain it
one time item: Purchases pulled forward ahead of expected price increases are inside the quarter's growth and may not recur.
line composition: Distributed Infrastructure revenue is derived, not reported, and Power and Storage are not separated.
other: The filing credits Power11 and differentiated storage products (claim c30).
other: The CEO names the growth of enterprise data beside AI adoption as the driver of the same growth (claim c26), with no split.
Quotes
“We see this as an increasingly important growth vector for IBM, driven by AI adoption and the rapid growth of enterprise data. We have been investing across Power and Storage, AI infrastructure to position ourselves for this market opportunity.”
“We are gaining share in Storage through differentiated offerings across flash, fusion, and tape, including AI-enabled capabilities that help clients scale and manage data for AI.”
“With clients prioritizing infrastructure investments, Distributed Infrastructure had its best performance in reported history, up 37 percent with strong growth in Power and Storage, and a backlog of approximately $500 million exiting the quarter”
“Storage revenue growth reflects our differentiated offerings including those with AI-enabled capabilities that help clients scale and manage data for AI. Power revenue growth was driven by continued demand for Power11 with its value proposition of resiliency, performance and Linux modernization.”
“Adding to Arvind's comments on the strength in Distributed Infrastructure, we exited the quarter with approximately $500 million of backlog, our highest on record, supporting continued momentum.”
Q1 2026described · described, no size · exploratory
Q2 2026direction only · described, no size · unknown
product revenue · cheap to verify
Lightwell: subscription for AI-remediated open-source packages
Not sized
The money had not started in the quarter: general availability was announced on 2026-07-08, after the quarter ended on 2026-06-30 (claims c56, c57), and management speaks of client sign-ups in the future tense (claim c53). A channel whose money has not started gets no ballpark (the methodology's rule on money not started, rules sweep of 2026-10-06). The former estimate, ibm-2026-cq2-f83, was nothing at the point with an allowance for early adopters at the high end.
described, no sizedisclosure: described· motive: exploratory· before LLMs: expanded
Lightwell was launched in the quarter after a new AI model sped up the discovery of vulnerabilities (claim c51): a subscription at per client per year for open-source packages patched or validated with AI models, with more than package versions available in its first weeks (claims c52, c35). The letter names early adopters without counting them, and management says it plans to measure the product by clients signed (claim c53). General availability came on 2026-07-08, after the quarter ended (claim c56), so the quarter's reading is unsized; the channel is registered now because the price and the launch are in this quarter's sources.
Evidence: 10 quotes, 3 figures, 2 confounds, 2 from before coverage
A subscription, launched after a new AI model sped up the discovery of software vulnerabilities, that gives clients open-source packages the company has patched or validated using AI models and its own engineers. It is priced per client per year. The work it automates, patching code and running tests against it, is cheap to check.
Why this motive
A separately priced AI-built product, per client per year (claim c52), would read offensive; it became generally available after the quarter ended and has no revenue yet, so the less durable motive is taken.
Before LLMs: expanded
Red Hat already sold enterprise open-source subscriptions at the anchor (Red Hat Enterprise Linux, OpenShift and Ansible), with revenue growth of in FY2024. Remediating open-source packages outside its own products, with AI models doing work that needed specialist engineers, is the part that was not there. The size is the change AI made, not the whole line.
“Red Hat: provides enterprise open-source solutions, for hybrid, multi-cloud environments, which includes Red Hat Enterprise Linux (RHEL), OpenShift, our hybrid cloud platform, as well as Ansible.”
“Cybersecurity risk to the company and its customers also depends on factors such as the actions, practices and investments of customers, contractors, business partners, vendors, the open source community and other third parties, including, for example, providing and implementing patches to address vulnerabilities.”
“The Mythos release in early April has accelerated the discovery of security vulnerabilities for clients. This creates a multi-billion-dollar TAM for IBM and Red Hat to help clients secure their open source software through our new capability, Lightwell.”
“Clients can subscribe to Lightwell for $1 million per year to access open source packages that have been remediated or validated. In the first two weeks of availability, we have already made more than 7,500 package versions available.”
“The same tools that are being used to attack all code, both proprietary and open source, can also be used to effectively be those experts and build those out.”
“Lightwell is a $5 billion commitment backed by new frontier AI capabilities and a global force of more than 20,000 engineers creating a trusted enterprise clearinghouse to address open source software vulnerabilities.”
“In the second quarter, we launched Lightwell, an investment to establish a trusted enterprise clearinghouse, combined with a global workforce of engineers, to address open-source software vulnerabilities at scale, which became generally available in July.”
“Lightwell, a new capability to address open source security vulnerabilities, leverages IBM and Red Hat's trust within the open source community, unique approach to AI, and global scale.”
“In aggregate, we believe that this is a multiple billion-dollar opportunity, which we're going to go after really fast and hard, leveraging expertise in AI and open source both.”
Cost imposed, or revenue lost, by others’ AI1 channel · 1 not sized
pricing packaging
Application software revenue exposed to agents replacing its users
Not sized
The quarter's sources say nothing about agents replacing the users of application software, and a silent quarter gives nothing to size; the Q1 estimate is not carried into a quarter with no statement on the channel.
inscrutabledisclosure: not mentioned· motive: imposed· before LLMs: new
Neither call, releases nor filing returns to agents replacing the users of application software or to the share of the portfolio exposed to them; the Q1 bound of stands unrepeated. The one related passage, the CEO's statement that value will increasingly shift toward the orchestration and data layers, says where the company sells and names no loss in the interaction layer, so it is read under the software channels, where it is cited, and not here.
Evidence: 0 quotes
Revenue the company could lose because of someone else's AI: agents doing the work people did in application software, which makes the interaction layer less valuable. Management names the mechanism and bounds its own exposure as a small share of the software portfolio.
Why this motive
A toll channel, imposed by construction. The quarter does not speak on agents replacing users of application software, so the motive is carried.
Before LLMs: new
The anchor does not discuss agents replacing the people who use application software, or an exposed share of the portfolio. Software revenue was in FY2024.
Q2 2026. Growing slower than revenue: selling, general and administrative (−0.9%), interest expense (−4.7%). A displaced cost shows up as a line that stays under the dashed revenue line. These are the audited lines, as first reported; nothing here is attributed to AI by the filing. Not drawn: restructuring charges, which one-time items move by more than 60% in a quarter; the values are in the table below.
Total costSelling, general and administrativeResearch and developmentInterest expenseRevenue