IBM (IBM)

IBM is the enterprise IT giant reinvented under Krishna into a software-led business. Software (44.4% of revenue, +10.6%, gross margin 83.5% — Red Hat, watsonx, ARR US$23,600 million) is a quality engine with a wide and stable moat — segment margin expanding across all four business lines in 2025, Consulting backlog rising with a book-to-bill of 1.03, IBM Z processing, per the company itself, 70% of the world's transactions. Consulting (31.2%) and Infrastructure (23.3%, cyclical, driven by the z17) dilute the consolidated figure to more modest growth. The dividend contributes a yield of 2.9%. The question the market began to raise months after this filing's close — whether artificial intelligence strengthens or cannibalizes the business — is not resolved in these sections of the 10-K.

A mature software business is valued on EV/owner-FCF net of stock-based compensation. IBM's free cash flow (its own measure, which treats financing receivables as investment) net of stock-based compensation yields owner-FCF; with corporate net debt (~US$31,700 million, excluding the self-funded Financing-segment debt of US$15,093 million) in the bridge, EV/owner-FCF is ~22×. The '15.5× cheap cash flow' shown by market aggregators is misleading twice over: it adds back stock-based compensation and the price ignores debt.

The base case projects owner-FCF growing toward an exit multiple of 16× — the low end of the quality-software band, disciplined by modest consolidated growth (high-growth software diluted by Consulting and cyclical Infrastructure) and by the AI-disruption risk the market began pricing in after this filing's close. That yields ~$213/share → a total return of +2%/year, of which the dividend (yield 2.9%) is a central part.

No margin of safety: at this price capital is preserved, but it is not bought below its value. The ~25% plunge of Jul-2026 corrected the overvaluation IBM had reached at the peak of AI enthusiasm toward a more reasonable value (~22×) — but it didn't create a bargain: at ~$232, IBM trades at a full multiple for its modest consolidated growth and its disruption risk. The verdict is Preserves value: a business with margins expanding across all four lines during 2025 and a wide moat this filing documents as intact, at a fair price, where the dividend carries a large share of the return and multiple compression offsets the growth. The adverse scenario (AI cannibalizes software and consulting, the mainframe declines, the IT multiple compresses) takes the value well below; the favorable one (watsonx and Red Hat accelerate, the mainframe recovers, the multiple re-rates) is a real upside. The asymmetry isn't especially favorable, and post-filing uncertainty is high.

Three things, with one caveat about timing. The caveat: this 10-K/Annual Report closes Dec-31-2025 and was filed Feb-24-2026 — it does not document the pre-announcement or the Jul-2026 plunge; this re-read confirms the business's quality through the fiscal year's close, not its performance in the second quarter of 2026.

What to watch: whether artificial intelligence strengthens IBM (watsonx and Red Hat capturing AI spend on the captive installed base) or cannibalizes it (hyperscalers capturing AI-infrastructure spend instead of traditional software and consulting) — the 10-K itself frames the AI risk as reputational and execution-related, not as direct cannibalization of Consulting, so the post-filing market narrative deserves to be tested against what the company actually reports. The mainframe cycle — the z17 drove IBM Z revenue +51.7% in 2025; if that cycle exhausts faster than expected, it drags down Infrastructure and the captive transaction-processing software. And the debt covenants — consolidated net interest expense coverage cannot fall below 2.20 to 1.0, with cross-default triggers on any other debt in default of at least US$500 million, while total debt keeps rising with M&A (HashiCorp closed in 2025, Confluent pending for mid-2026).