Enterprises are sweating legacy IT assets as AI investment grows

Ensono's survey of 500 US and UK IT leaders found 78% view legacy systems as more important than two years ago, with AI investment driving companies to extend rather than replace mainframes and other existing infrastructure. While 45% are scaling AI organization-wide and 44% are funding targeted use cases, integration challenges (33%) and infrastructure constraints (28%) remain key barriers. More than half of organizations are optimizing legacy platforms in place, while enterprise IT asset-refresh cycles are reportedly extending from roughly five to seven years.

Analysis

The investable implication is a shift in enterprise IT spend from wholesale infrastructure replacement toward integration, operations, data access, and application-layer modernization. KD is the clearest listed beneficiary: longer-lived heterogeneous estates expand the addressable managed-services pool and make recurring run-and-transform contracts stickier, although revenue conversion will lag survey intent by 2-4 quarters. IBM is an unlisted-data omission worth monitoring alongside KD; its mainframe software, consulting, and hybrid-cloud positioning can monetize both retention and selective modernization without requiring customers to fund a disruptive migration.

AVGO faces a more nuanced risk than simple VMware license dissatisfaction: higher virtualization costs can accelerate workload repatriation to mainframes, public cloud, or alternative hypervisors, reducing VMware seat growth and increasing renewal friction. HPE is exposed to a delayed server/storage refresh cycle, with the adverse mix concentrated in general-purpose enterprise infrastructure rather than AI systems; this can pressure near-term revenue even if customers ultimately spend more on targeted AI deployments. The second-order winner is Gartner (IT): uncertain architecture choices, licensing changes, and AI governance raise demand for advisory seats, a relatively asset-light spend item that is easier to approve than a multiyear transformation program.

Consensus may overestimate the near-term hardware uplift from enterprise AI. If AI projects remain constrained by workflow integration and data readiness, customers will first purchase services, observability, security, and data-access tooling rather than broad compute refreshes; the revenue pool shifts toward incumbents managing legacy complexity. The thesis is falsified if enterprise CIO surveys or company bookings show refresh cycles shortening, VMware renewal rates stabilizing without elevated concessions, or HPE reporting a material reacceleration in non-AI server orders over the next two earnings cycles.

Key Decisions for Investors

• Initiate a 3-6 month long KD / short HPE pair, sized modestly: KD should benefit from deferred replacement and modernization-in-place, while HPE remains exposed to elongated general infrastructure refresh cycles. Reassess if HPE's non-AI compute orders or backlog conversion inflect materially upward; target 10-15% relative return with a 5-7% pair stop.
• Maintain an underweight or tactical short AVGO into VMware renewal datapoints over the next 1-2 quarters, preferably paired against IBM or KD to isolate licensing-driven workload substitution from broad AI beta. Cover if reported VMware renewals, net retention, or pricing realization demonstrate customers are accepting the new model without increased migration activity.
• Add IT on weakness for a 6-12 month horizon rather than chase AI hardware proxies. Advisory demand should be a leading indicator of delayed architecture decisions; require evidence in contract value growth and enterprise-tech consulting demand before increasing size.
• Set an earnings watch on KD for bookings, contract duration, and margin commentary: a rise in modernization and platform-management signings without corresponding labor-cost pressure would validate the thesis and justify upgrading the position from a pair-trade leg to a standalone long.