IBM (IBM) Q3 2026 Earnings Call Transcript

IBM reported third-quarter revenue of $17.1 billion, up 6% at constant currency, and non-GAAP EPS of $2.95, up 11% year over year; software revenue grew 9% and infrastructure 5%. Management maintained full-year revenue growth guidance of 4%–5% and raised its free-cash-flow growth expectation to more than $1 billion year over year. First-nine-month free cash flow was about $8 billion, up 8%, while IBM returned $4.8 billion in dividends year to date.

Analysis

The useful signal is not the quarter’s headline growth but improving software demand quality: management attributed most of the sequential acceleration to net-new business, while roughly one-quarter came from delayed deals. That supports a better forward setup than a pure catch-up quarter, but the double-digit transaction-processing print is unusually timing-sensitive; management itself expects moderation. Avoid extrapolating it into a smooth mainframe growth curve.

IBM’s hybrid/sovereign-AI proposition is a portfolio complement, not evidence that it captures AI compute economics. More private deployments could lift demand for infrastructure and governance tools while shifting some workloads away from hosted frontier-model APIs. NVIDIA may benefit from on-prem GPU demand; Alphabet could face conditional substitution pressure if cheaper open-weight models prove adequate for routine workloads. IBM’s ability to monetize that shift depends on customer adoption and cross-selling, not management’s market narrative alone.

Near term, stronger FCF guidance and margin execution improve earnings quality, but acquisition spending, elevated debt, and reinvestment temper the cash-return story. Over 1–3 months, the key test is Q4 conversion of bookings and backlog; January guidance will clarify whether software growth is broadening or the quarter was partly timing. Quantum is a long-dated option, not a basis for this earnings trade. Without valuation and post-call price data, the signal supports buying weakness rather than chasing an assumed rerating.

Key Decisions for Investors

• IBM: Favor a staged long on post-earnings weakness rather than an opening-gap chase. Thesis: recurring software demand plus productivity can sustain earnings growth beyond the mainframe catch-up. Reassess if Red Hat bookings soften, software growth decelerates materially, or full-year FCF delivery weakens; check valuation and price action before sizing.
• Treat IBM’s transaction-processing rebound as a catalyst with limited duration: monitor Q4 TP results and January 2027 guidance. A return toward flat growth would not by itself break the thesis; renewed declines alongside weaker installed MIPS or software bookings would.
• Watch for a relative beneficiary trade in NVIDIA if enterprise on-prem and sovereign-AI deployments translate into GPU orders. This is a conditional demand hypothesis, not confirmation from IBM’s call; verify independent infrastructure spending and NVIDIA data-center demand before initiating.
• Track Alphabet as a potential relative laggard only if customers demonstrably shift routine inference from hosted frontier APIs to open-weight private deployments. Do not short on this call alone: evidence of token-cost substitution, rather than experimentation, is the required catalyst.