IBM — Q2 2026 Financial Report Analysis

A quarter lost in the last two weeks of June

IBM grew revenue 1.1% to $17.16 billion in the three months to 30 June 2026 — and management said plainly that this was not the plan. The 10-Q's own words: "our second-quarter performance was below our expectations." The miss was not spread across the quarter or the portfolio. It was concentrated in IBM's mainframe franchise and it happened late: "In the final weeks of June, we saw a shift in client spending priorities that resulted in numerous large deals failing to close within the expected timelines, driving the majority of the shortfall."

The reason those deals slipped is unusually specific, and it matters for reading the rest of the quarter. Customers redirected budget toward "servers, storage, and memory purchases to secure supply-constrained infrastructure ahead of expected price increases." In other words, enterprises facing a tightening market for AI-adjacent hardware bought that hardware first and postponed signing large mainframe and mainframe-software contracts. IBM saw both sides of that trade inside its own results: IBM Z revenue fell 42.0%, while Distributed Infrastructure (Power servers and Storage) rose 37.3% — its "strongest quarter on record."

The numbers

A note on two terms used throughout. Constant currency ("cc") strips out the effect of exchange-rate moves, so you see whether a business actually sold more, not whether the dollar happened to move in its favour. Margin is the share of revenue left over after a given set of costs; gross margin is what remains after the direct cost of delivering the product or service, before research, sales and administration. IBM's income statement has no operating-income subtotal — interest expense and other income sit inside one combined "total expense and other (income)" line — so pre-tax margin is the closest equivalent, and it is the figure shown as "operating margin" in the summary strip above this analysis.

Where the profit went, and what rescued the bottom line

Gross profit fell 0.7% on revenue that rose 1.1%, and gross margin gave up a full point to 57.7%. The filing attributes this to "our revenue shortfall and mix, partially offset by productivity actions" — meaning the high-margin mainframe software that did not close was replaced in the revenue mix by lower-margin distributed hardware. That is a mix effect, not a pricing problem, and it can reverse if the deferred deals close.

Below the gross line, the story is investment. Research and development rose 10.2% to $2,311 million while selling, general and administrative expense fell 0.9% to $4,981 million — IBM is funding R&D out of sales-and-admin productivity rather than out of growth. Two smaller items cut in opposite directions: intellectual property and custom development income fell to $166 million from $215 million (a headwind, since it is booked as a credit against expense), while other income rose to $185 million from $39 million (a tailwind). Net of everything, total expense rose only 0.7%, but with gross profit down, pre-tax income fell 4.5%.

Net income fell only 1.3%, and the gap between those two declines is tax. The provision was $313 million against $404 million a year earlier — an effective tax rate of 12.6% versus 15.6%. A lower tax rate, not operating performance, is what kept the bottom line roughly flat. Diluted EPS fell slightly more than net income (−1.7% vs −1.3%) because the diluted share count rose 0.6% to 953.3 million: IBM has not repurchased shares since the Red Hat deal in 2019, and it issued and assumed 3.0 million stock awards worth $665 million in the Confluent acquisition.

GAAP down, adjusted up — and why the gap is the Confluent bill

The single widest divergence in this quarter is that GAAP EPS fell 1.7% while IBM's own operating (non-GAAP) EPS rose 4.6%, to $2.93. The bridge between them is almost entirely acquisition accounting: after-tax acquisition-related charges of $548 million, up 23.8% from $443 million, plus $76 million of non-operating retirement-related costs (versus $17 million).

Those acquisition charges are mostly the cost of Confluent, the data-streaming company IBM bought in the first quarter of 2026 for $11.6 billion in total consideration ($11.5 billion of it cash), which added $3,834 million of acquired intangible assets and $7,149 million of goodwill to the Software segment. Total intangible amortisation reached $817 million for the quarter, against $687 million a year earlier.

This is worth stating directly, because the two figures point in opposite directions for a reason. Confluent is a meaningful part of the growth the adjusted number is celebrating — the filing says Data revenue rose 18.9% "reflecting the contribution from recent acquisitions, primarily Confluent," and that "recent acquisitions delivered a strong contribution to Software revenue growth." The adjusted EPS figure counts Confluent's revenue while excluding the amortisation of what IBM paid for it. That is a standard and permitted presentation, but a reader comparing the 4.6% adjusted EPS growth against last year should know that part of the numerator was purchased and the purchase price is not in the denominator of the comparison.

Segments

Software: $7,761M, +5.1% (+4.6% cc). Segment profit $2,502M, +8.9%; margin 32.2%, +1.1 pt. Growth came from the recurring base — roughly 80% of Software revenue, which "delivered healthy growth" — and from acquisitions. Hybrid Cloud (Red Hat) rose 11.2% on subscriptions and "stabilization in consumption-based services," with OpenShift annual recurring revenue now above $2 billion. Data rose 18.9%, mostly Confluent. Automation rose 3.6%. Against that, Transaction Processing — the mainframe software stack — fell 8.1%, directly tied to the IBM Z shortfall. The filing explains why the two move together: the remaining ~20% transactional slice of Software is "predominantly related to the incumbency with IBM Z and the associated enterprise license agreements." When a mainframe deal slips, the software licence attached to it slips with it. Segment ARR reached $24.6 billion, up about $2 billion year on year, with Confluent a large part of that step-up. Note that Software gross margin fell 1.3 points to 82.6% even as segment profit margin rose 1.1 points — cost discipline below the gross line more than offset the mix.

Consulting: $5,327M, +0.2% as reported, +1.1% cc. Segment profit $647M, +15.1%; margin 12.1%, +1.6 pt. Revenue was flat, but this was the most profitable-looking part of the quarter: gross margin rose 1.4 points to 28.9% and segment profit grew fifteen times faster than revenue. Demand is described as coming from "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." Forward indicators held up: signings rose 5.0% (+5.9% cc) — the second consecutive quarter of growth — trailing-twelve-month book-to-bill was about 1.05 (i.e. IBM signed slightly more new work than it billed, so backlog is still building), and backlog stood at $30.8 billion at quarter end.

Infrastructure: $3,835M, −7.4%. Segment profit $835M, −13.4%; margin 21.8%, −1.5 pt. Two things happened at once here, and separating them matters. First, a genuine comparison effect: the z17 mainframe launched in June 2025, so Q2 2026 was always going to be measured against "the strongest start to a mainframe program in our history," and IBM says it "expected declines year to year in revenue." Second, a real miss on top of that comp: results "were worse than our expectations." IBM Z fell 42.0%. Distributed Infrastructure rose 37.3% with double-digit growth in both Storage and Power — Storage on AI-related data capabilities, Power on demand for Power11 — and the press release notes Power and Storage have built an order backlog of nearly $500 million. The segment's gross margin loss of 3.1 points is the mainframe-to-distributed mix showing up in the P&L.

Geography. Americas revenue fell 0.5% as reported but 1.0% in constant currency; EMEA rose 4.1% reported but only 2.1% constant currency; Asia Pacific rose 0.3% reported and 5.2% constant currency. Asia Pacific is the one to notice: a roughly five-point currency headwind turned real regional growth of 5.2% into almost nothing in dollars. At the total-company level currency was neutral — reported and constant-currency growth were both 1.1%.

Takeaway: The 42% collapse in IBM Z is a timing and comparison event, not evidence the mainframe franchise is breaking — the deals slipped in the last fortnight of June because customers rushed to buy supply-constrained servers, storage and memory first, and IBM booked the other side of that same trade as record 37% Distributed Infrastructure growth. What is not a timing event is the guidance cut it forced, and the fact that half the year is gone with free cash flow flat while the full-year target still calls for a $1 billion increase.

Cash: better than it looks in the quarter, harder than it looks for the year

Operating cash flow was $2.6 billion, up $0.9 billion year on year — but free cash flow was $2.5 billion, down $0.3 billion. The two diverge because IBM's free cash flow definition removes the change in financing receivables (loans IBM extends to customers to buy its own equipment), and that line swung by $1.2 billion in the quarter. The operating-cash improvement is largely that swing; underlying free cash generation was slightly weaker than a year ago.

For the first half: operating cash flow $7.8 billion (up $1.7 billion) and free cash flow $4.8 billion — exactly flat year on year. IBM still expects full-year free cash flow to rise by about $1 billion. With none of that increase delivered in six months, the entire improvement has to come in the second half.

The balance sheet absorbed the acquisitions. Cash, restricted cash and marketable securities ended the quarter at $8.2 billion, down $6.3 billion from year-end 2025, after $10.5 billion spent on acquisitions year to date. Debt, including $13.0 billion of IBM Financing debt, totalled $62.0 billion, up $0.7 billion year to date. IBM returned $1.6 billion to shareholders in dividends in the quarter; there is still no buyback.

Guidance: cut, and by more than it first appears

IBM lowered its full-year outlook, and the size of the cut is clearer when the two statements are put side by side. In April, with Q1 results, the company "continues to expect full-year constant currency revenue growth of more than 5 percent," with currency "about a half-point to one-point tailwind to growth for the year." In July: "The company now expects full-year constant currency revenue growth in the range of four-to-five percent. At current foreign exchange rates, currency is expected to be neutral to growth for the year."

So the underlying growth target came down from above 5% to 4–5%, and the expected currency benefit went from up to a point to nothing. In actual dollars, the revenue expectation dropped by roughly one and a half to two and a half points. Free cash flow guidance was unchanged at about +$1 billion year on year, and management now expects "improved pre-tax income margin expansion for the full year" — profitability guidance went up while revenue guidance went down, which is consistent with the cost pattern in the quarter (SG&A down, productivity actions cited repeatedly).

What to watch

The central question for the second half is whether the deals that failed to close in June actually close, or whether they were lost. The filing's language leans toward the former — "near-term client buying dynamics that we will work through" and deals "failing to close within the expected timelines" rather than being cancelled — and the structure of the business supports that reading: mainframe customers signing enterprise licence agreements are not casual buyers who walk away. But the claim is unverifiable until Q3, and it is doing a lot of work in the full-year guidance, which still implies acceleration.

Three other things carry into the second half. IBM signed a definitive agreement on 22 July 2026 to acquire an unnamed business that will be integrated into the Infrastructure segment, expected to close in the second half subject to regulatory approval — another acquisition, and therefore more of the GAAP-versus-adjusted divergence described above. Quantum spending is now an explicit commitment of more than $10 billion over five years, including a letter of intent with the U.S. Department of Commerce to build a quantum wafer foundry, with the first large-scale fault-tolerant machine targeted for 2029 — a long-dated investment that consumes cash well before it produces revenue. And the free-cash-flow target requires a second half meaningfully better than the first.

On balance: the recurring engine is intact — 80% of Software recurring and growing, ARR at $24.6 billion, Consulting backlog at $30.8 billion with book-to-bill above 1 — and margins in the two larger segments improved year on year. The weakness is concentrated in the most cyclical, most lumpy part of the portfolio at the worst point in its product cycle. That is a recoverable setup, but IBM has now spent its margin of safety for the year: the guidance assumes the slipped deals come back, and there is no second cushion if they do not.

All figures from IBM's Form 10-Q for the quarterly period ended 30 June 2026 (filed 23 July 2026), except full-year expectations, cash-and-debt balances and the Power/Storage order backlog, which are from Exhibit 99.1 of the Form 8-K furnished 22 July 2026, and the prior full-year outlook, which is from Exhibit 99.1 of the Form 8-K furnished 22 April 2026. Operating (non-GAAP) earnings are IBM's own measure, excluding acquisition-related charges, intangible amortisation and non-operating retirement costs.