IBM says AI is insane in the mainframe as z17 sales surge

Steven

IBM says AI is insane in the mainframe as z17 sales surge

Summary

IBM reported strong FY results driven by infrastructure and mainframe demand as the company leans on generative AI to modernise legacy systems. Q4 revenue from continuing operations reached $19.7bn, up 12% year‑on‑year, and net income rose 91% to $5.6bn. Full‑year revenue hit $67.5bn (up 8%) and net income was $10.6bn (up 76%).

The firm says its cumulative GenAI book of business now exceeds $12.5bn, split roughly between Software and Consulting, and it has achieved $4.5bn of annual run‑rate productivity savings. IBM emphasised a record launch and strong sales for its z17 mainframe series — IBM Z performance grew 48% year‑on‑year, the highest Z revenue in about 20 years. IBM is positioning GenAI (via tools such as Watson Code Assistant for Z) to automate COBOL refactoring and enable sub‑millisecond, in‑line AI with transactions. The company also warned that AI‑driven demand is keeping DRAM and server pricing elevated.

Key Points

  • IBM Q4 revenue from continuing operations: $19.7bn (+12% YoY); net income $5.6bn (+91% YoY).
  • Full‑year revenue $67.5bn (+8%); net income $10.6bn (+76%).
  • GenAI book of business > $12.5bn; Software > $2bn, Consulting > $10.5bn, both saw strong quarterly growth.
  • IBM Z (z17) revenue surged: Z performance up 48% year‑on‑year, highest annual Z revenue in ~20 years.
  • Watson Code Assistant for Z aims to refactor COBOL to Java or modernise code, lowering the barrier to modernisation.
  • In‑line mainframe AI offers millisecond latency vs multiple seconds when off‑platform — a key argument for on‑prem AI for transactional workloads.
  • Supply pressures: AI server demand is sustaining higher DRAM and server prices through the year.

Context and Relevance

IBM’s results show a notable intersection of legacy enterprise computing and modern AI. Mainframes are being marketed not merely as legacy replacements but as competitive, sovereign, low‑unit‑cost platforms for latency‑sensitive AI workloads. The z17 launch performance indicates a renewed commercial appetite for on‑prem infrastructure as organisations weigh sovereignty, cost per transaction and latency when deploying generative AI. At the same time, component shortages and price pressure around DRAM and servers demonstrate how AI demand ripples across the wider hardware market.

Why should I read this?

Short version: mainframes aren’t dead — IBM’s making them relevant for AI. If you care about enterprise AI, legacy modernisation, or on‑prem sovereignty, this is a useful snapshot of where money and momentum are heading. The piece shows how AI tooling is being used to fix COBOL headaches and why that matters for latency and cost. Worth a skim if you run or advise large, transactional systems.

Author style

Punchy — the article cuts straight to the point: IBM is monetising AI while reviving Z hardware sales. Read the detail if you want the figures and the rationale behind IBM’s mainframe comeback; otherwise, the summary gives you the business impact in one go.

Source

Source: https://go.theregister.com/feed/www.theregister.com/2026/01/29/ibm_q4_2025/