According to Zhitong Finance APP, after the close of U.S. markets on Wednesday, IBM (IBM.US) reported its third-quarter 2026 financial results. The data showed that the company’s revenue for the quarter was $17.103 billion, a 5% year-over-year increase, or a 6% increase on a constant currency basis; GAAP net income from continuing operations was $2.199 billion, up 26% year-over-year, with diluted earnings per share of $2.31, also up 26% year-over-year. Non-GAAP adjusted net income from continuing operations was $2.812 billion, up 12% year-over-year, with adjusted earnings per share of $2.95, up 11% year-over-year. IBM’s stock rose more than 2% in after-hours trading.
In terms of profitability, IBM's GAAP gross margin for the third quarter was 56.8%, a decrease of 0.5 percentage points year-over-year; the non-GAAP adjusted gross margin was 58.5%, a decrease of 0.2 percentage points year-over-year. Adjusted pre-tax profit was $3.295 billion, an increase of approximately 9% year-over-year, with an adjusted pre-tax profit margin of 19.3%, up 0.7 percentage points year-over-year.
Looking at the business segments, software revenue increased 8% year-over-year to $7.804 billion, including Red
Hat's hybrid cloud business grew 11%, data business grew 19%, and transaction processing business grew 9%, but automation business revenue declined 3%. Consulting business revenue was $5.335 billion, essentially flat compared to the same period last year; infrastructure business revenue increased 4% year-over-year to $3.707 billion, with distributed infrastructure business growing 31%. IBM
Mainframe business decreased by 20%.
In terms of cash flow, IBM's operating cash flow for the third quarter was $4.025 billion, an increase of approximately $900 million year-over-year; free cash flow was $3.027 billion, up approximately $700 million year-over-year. For the first nine months of the year, the company's cumulative free cash flow reached $7.787 billion. As of the end of the third quarter, IBM held $11.1 billion in cash, restricted cash, and marketable securities, with total debt of $63.8 billion.
For the full year 2026, IBM maintains its expectation of 4% to 5% revenue growth on a constant currency basis and anticipates that currency exchange rates will have a broadly neutral impact on full-year revenue growth. Meanwhile, the company has raised its free cash flow outlook, expecting full-year free cash flow to increase by more than $1 billion compared to 2025.
Below is the full transcript of IBM's earnings call (translated with the assistance of an AI tool):
I. Executive Speech
Chairman and CEO Arvind Krishna
Revenue growth accelerated to 6% in the third quarter, exceeding the baseline scenario set in July. This performance reflects accelerated growth across all business segments: software up 9%, infrastructure up
5%, with consultative improvements. Disciplined execution also drove strong performance in profits and free cash flow.
Software segment: Red Hat maintained an 11% growth rate, with order growth returning to high double digits; OpenShift's annual recurring revenue (ARR) growth accelerated to 30%.
Above. Transaction processing (TP) increased by 10%, confirming that the weakness in Q2 was a timing issue rather than a demand issue. Data business grew by 20%, with Confluent continuing to perform strongly. Infrastructure growth
5%, with distributed infrastructure growing by 32%.
AI and Storage: AI continues to drive demand for storage, and the company is expanding its market share through differentiated solutions in Flash, Fusion, tape, and IBM Z storage. Consulting revenue has increased.
2% of customers are modernizing to securely and scalably deploy AI.
Z17 host: Continuing the platform's strongest innovation cycle, delivering over $1 billion in incremental revenue across six quarters, benefiting from advantages in security, reliability, and total cost of ownership (2 to 15 times lower than non-platform solutions).
The Z system, launched in July as a single-rack/rack-mounted model, generated the strongest mid-range product launch revenue in nearly 20 years this quarter.
Lightwell: This is a multi-billion-dollar opportunity in open-source software security, with a pipeline nearing $700 million and over 40,000 patch releases published to the web.
AI Strategy: Enterprise AI is entering a new phase—deploying more models or agents is no longer enough; the key lies in safely, efficiently, and sustainably running AI within the enterprise. IBM
Complete capability integration across four platforms: Confluent provides the real-time data foundation, and Watsonx (including Bob and Orchestrate) enables AI.
Driven development and agent orchestration, HashiCorp and Apptio enable identity, access, and policy automation, while Red Hat provides an open hybrid architecture. Data is at the core of AI operational models, with 72%
Enterprises report insufficient real-time data infrastructure. To address this, we are introducing Confluent’s Streamhouse capabilities to continuously capture, transform, and query real-time data, unifying operational and analytical data.
Cost and Governance: Enterprises are most concerned with "how much it costs and who controls it." Apptio has connected AI spending to business outcomes; developed in collaboration with Red Hat and Google, LLM-D
Reduce large model operational costs by 5 to 10 times; Bob now supports local deployment. At the governance level, HashiCorp Vault can register, authorize, audit, and revoke agents.
Identity is collaborating with NVIDIA's Open Agent Safety Platform. We believe that "agent identity" will become as fundamental to enterprise AI as "internet identity" has been.
the foundation.
Quantum computing: In July, collaborated with the University of Chicago, Algorithmiq, and Qedma to achieve three independent demonstrations of quantum advantage. The Nighthawk r2 processor reached 7,500.
Gate, executing 100,000 circuits per second, with performance improved by up to 19 times over the previous generation. IBM has been selected for Phase Three of the U.S. Defense Advanced Research Projects Agency (DARPA) Quantum Benchmarking Program; 16 quantum business agreements have been signed this year.
Cross-industry corporate collaborations, including AT&T and General Motors. The first modular cryogenic system has been successfully connected, and the 300,000-square-foot basepisc manufacturing facility is progressing, along with a $10 million grant from the Department of Commerce.
Hundreds of millions of dollars in semiconductor manufacturing grants give the company confidence in scaling up.
Full-year outlook: We reaffirm our revenue growth forecast of 4% to 5% for 2026 and raise our full-year free cash flow growth expectation to over $1 billion.
CFO James Kavanaugh: Financial Data and Outlook
Revenue increased 6% in the third quarter, operating pre-tax profit margin expanded by 70 basis points, and diluted operating earnings per share rose 11%. Free cash flow for the first nine months totaled approximately $8 billion, an 8% year-over-year increase.
Software: Revenue accelerated by 4 percentage points to 9% (including over 4 percentage points of organic growth), driven by high-value annual recurring revenue (ARR), which increased by 8% to $249M.
Billion U.S. dollars. Three of the four software sub-segments achieved double-digit growth and successfully closed approximately two-thirds of the second-quarter slipped deals. Red Hat grew by 11%, and OpenShift has reached $2.4 billion in annual recurring revenue (ARR).
Business growth exceeds 30%; virtualization has generated nearly $800 million in contracts and a pipeline of approximately $750 million since early 2024. Data growth of 20% (Confluent orders up approximately 20%); TP
Grew 10%; HashiCorp experienced order growth across regions, but automation was held back by a high base.
Infrastructure: Growth accelerated to 5%, distributed infrastructure grew 32%, and storage achieved record growth while continuing to expand its market share. The z17 delivered exceptional performance, marking the strongest mid-range launch in nearly 20 years. Approximately 85% installed.
Host customers with MIPS capacity are maintaining or expanding their capacity, resulting in cumulative installed capacity doubling compared to previous cycles.
Consulting: Revenue grew 2%, strategy and technology grew 4%, and cybersecurity saw nearly double-digit growth. Generative AI accounted for approximately 30% of consulting revenue and 35% of backlog for the quarter. We achieved this through our partnership with OpenAI
New strategic partnership, deepening the relationship with Google Cloud, and expanding the customer capabilities ecosystem.
Profitability and Cash Flow: Adjusted EBITDA and operating pre-tax margin expanded by 60 and 70 basis points, respectively; software and infrastructure segment margins expanded by 210 and 120 basis points, respectively. First nine months
Free cash flow for the month was approximately $8 billion, an increase of about $600 million year-over-year, driven by an approximate $1 billion increase in adjusted EBITDA, partially offset by capital expenditures, cash taxes, and net interest expenses. Cash at period end: $11.1 billion.
$1.2 billion, with $4.8 billion in dividends distributed over the first nine months, and $63.8 billion in debt (including $12.7 billion in financing business, with nearly 80% of the receivables portfolio rated investment-grade).
Full-year guidance: Maintain revenue growth of 4% to 5%, raise free cash flow growth to over $1 billion. Software full-year growth of approximately 7% to 8% (baseline approximately
(7%), with double-digit growth in hybrid cloud and data, low single-digit growth in automation, flat TP; low single-digit growth in infrastructure; low to mid-single-digit growth in consulting. Full-year operating pre-tax profit margin expanded by 100 basis points.
One basis point; the operating tax rate remains in the mid-teens. Free cash flow realization rate has been approximately 130% for multiple consecutive years.
II. Q&A Session
Amit Daryanani, analyst at Evercore ISI
Key question: How have customer spending priorities changed? Are infrastructure investments beginning to drive incremental software demand? Red Hat
Does order growth in double digits signal a software speedup by 2027?
Arvind Krishna: Customer investment in AI remains strong, but to realize value from AI, applications must be rewritten or modernized—directly benefiting Red Hat.
OpenShift—we are still at the beginning of this cycle. 90% of enterprise data resides on-premises rather than in the cloud; unlocking the value of this data is key.
Confluent. Automation only delivers benefits when applications are scaled and enterprises need to manage hundreds or thousands of applications. Regarding Red Hat, we are confident in double-digit growth for the fourth quarter, but we are not providing guidance for 2027.
Last year, it was stated that, generally, high order growth translates into subsequent revenue growth.
James Kavanaugh: The overall software performance was satisfactory, with all four points of sequential acceleration driven organically. High-value recurring revenue accounted for approximately 80%, growing in high single digits, while transactional business (the remainder
20%), rebounded into double digits, with Enterprise License Agreements (ELAs) growing over 20% and net revenue retention exceeding 130%. Approximately one-quarter of the software acceleration came from regained deals, while three-quarters came from net new business. TP
Double-digit growth confirms the strategic demand for host software, putting to rest the previous debate over "demand destruction versus timing issues." Six quarters after the launch of the z17, installed MIPS and shipped MIPS have both experienced high double-digit growth.
Growth exceeds 100%, representing a future monetization opportunity for TP acceleration.
Bank of America analyst Vamsi Mohan
Key question: What is the proportion of TP in large slippage orders? Is the annual range of 4% to 5% on the higher end? Can TP’s performance across cycles still be considered low single digits?
Arvind Krishna: We always believed that the concerns about decreased demand for mainframes last quarter were merely a matter of timing. As customers deploy hardware and MIPS and transaction volumes grow, demand for mainframe software will inevitably rebound. Previously, for TP
The modest to low single-digit decline is conservative, as the actual second-quarter deferrals, combined with the normal procurement cycle in the third quarter, are now being realized. I do not expect TP.
Dual-digit growth will persist but return to more normalized levels by the fourth quarter, possibly declining slightly. This is driven by underlying demand for hosts in terms of resilience, cost-efficiency, and an "eight nines" availability architecture.
James Kavanaugh: Software quarterly revenue was approximately $7.6 billion, with a shortfall of about $400 million in the second quarter. We closed approximately two-thirds of the slipped deals (over $250 million), accounting for about
25%, primarily concentrated in TP. In the fourth quarter, we cautiously anchored software growth at approximately 7%, resulting in full-year revenue of about $32 billion, in line with market expectations, which gives us confidence.
Ben Raizes, analyst at Melius Research
Key question: Do you still have confidence in achieving the software’s long-term goal of "near double-digit" growth? What role do acquisitions play in this?
James Kavanaugh: We are confident in our strategy and portfolio. Early signals in the software segment include accelerated high-value recurring revenue, full-year high single-digit growth, and fourth-quarter growth approaching double digits; on new innovations, Red
Hat accelerates to high double digits, Lightwell pipeline reaches $700 million (against a trillion-dollar open-source vulnerability market), virtualization and hybrid cloud containerization opportunities; 2026 is the year to integrate Confluent.
In 2027, it will be the year of growth synergy; on the TP side, installed MIPS will double in five years and increase tenfold in ten years, serving as a clear monetization vehicle. We will provide quantitative guidance in January.
Barclays analyst Remo Lengauer
Key questions: How is Confluent's upsell/cross-sell progressing? What are the drivers of gross margin?
Arvind Krishna: Confluent is responsible for moving data in real time and keeping data in place so it remains accessible to AI; complementing this is the lakehouse, which we are excited about coming soon.
Cross-selling opportunities brought by Streamhouse. Unlocking data and avoiding the additional costs of duplicating data for AI are core capabilities for enterprise AI.
James Kavanaugh: Over the past four years, the company's operating leverage improved by approximately 1,000 basis points, cash increased by about $7 billion, and free cash flow margin expanded by 700 basis points. The quarter delivered $500 million.
A total productivity gain of basis points, with approximately one-third allocated to protecting profit margins and two-thirds reinvested. We have cumulatively unlocked over $5.5 billion in productivity, creating a powerful flywheel for investment flexibility and competitiveness.
Jefferies analyst Brent Thill
Key issue: Some AI projects have failed to deliver on their initial promises—have you seen companies reassessing and returning?
Arvind Krishna: Large enterprises are increasingly scrutinizing the real return on AI spending, concerned about unchecked investments. The productivity advantages of AI are real, but many underestimate the challenges of implementing AI
The modifications required to integrate into workflows cannot be solved simply by installing an RPA tool. Companies are moving away from small-scale pilots and instead focusing on scaling a few select use cases, while choosing trusted software partners and managing token usage.
Cost. This is favorable for OpenShift, Confluent, and the host.
Citi analyst Fatima Bhanani
Key question: In the recovery of trading volume (TP), what is the contribution of volume versus price? Was pricing power utilized?
James Kavanaugh: TP is critical software running on the most enduring platform (host), which supports approximately 70% of global transaction value and 45 of the top 50 banks and the top 10
Seven of the largest retailers. We operate approximately 1.45 billion MIPS globally; MIPS shipments in the sixth quarter after the z17 launch were 125% higher than those of the z16 during the same period, with MIPS shipment growth exceeding
100%, installed MIPS increased by 18%. The former represents immediately monetizable consumption and capacity, while the latter reflects future market potential. Unit sales for z16 remained flat in its launch year, then accelerated annually thereafter; z17
We are currently assuming a flat outlook, but 100% of shipped MIPS versus 18% of installed MIPS indicate a significant monetization opportunity by 2027.
JPMorgan Chase analyst Brian Essex
Key issue: Customer focus on "sovereign AI" during AI transformation evaluation, and how to protect data and intellectual property.
Arvind Krishna: Sovereign AI is becoming an increasingly important issue—barely noticed in 2024, gaining attention from a few in 2025, and by 2026
The trend has clearly intensified this year. Using large public large models leaves usage traces in underlying logs, and companies want to ensure their data and logs are not accessed by outsiders. This has spurred the rise of local AI.
The dual surge in infrastructure and open-source weight models, as the latter grants enterprises the authority and intellectual property to run locally or on private clouds. A CEO of a major bank
Clearly state that while external front-end model experiments are permitted, internal operations must be sovereign. This will benefit infrastructure, semiconductor companies, and players that help businesses deploy AI in secure, private environments.
Matt Swanson, analyst at RBC Capital Markets
Key issue: Businesses balancing "AI costs" against "control," and how to design network architectures that accommodate both.
Arvind Krishna: By the end of 2024, the gap between state-of-the-art frontier models and open-source models was approximately 2:1 to 3:1—a disparity too large to be acceptable; by 2025
The year-end gap has narrowed and is expected to narrow further. The real challenge for enterprises lies in the lack of core engineers to optimize GPU clusters, network proximity, and result caching. We estimate that large enterprises need 2 to 3
A single rack can run open-source weight models, with token costs only 10% to 25% of those for state-of-the-art models, while achieving 90% of their performance. This is a calculation any business would make: 90% of the performance at 10% of the cost.
The cost. Most tasks—such as summarizing emails, reviewing contracts, and responding to customer service inquiries—can be accomplished with cheaper alternatives, which is the key driver of industry transformation.