AI is not killing ERP. Nor is it killing SaaS. But it is exposing how enterprises sequence decisions.

The AI debate in 2026 has become oddly binary.

Either white-collar work shrinks dramatically and SaaS margins collapse.

Or nothing much changes at all.

The more interesting story sits between those extremes.

In its February piece, Harvard Business Review argued that AI investments are not failing because the technology is weak. They are underperforming because organisations have not redesigned how work flows. Layoffs in some firms are outpacing realised productivity gains. “Workslop” appears when AI output requires human correction. Cultural dissonance grows when expectations accelerate faster than operational change.

That is the inside-the-firm view.

The outside view looks different.

Commentary in The Register dismissed talk of a “SaaS-pocalypse” - the idea that AI will destroy enterprise software economics. Enterprise incumbents such as SAP, Oracle and Salesforce are protected less by code scarcity than by data gravity, embedded processes and switching friction. Cheaper feature development does not remove integration risk, audit requirements or regulatory complexity.

Enterprise IT is slow-moving for structural reasons.

Add a third layer: capital discipline.

Higher interest rates have raised the cost of capital and compressed valuations. Multi-year ERP transformations - particularly migrations from SAP ECC to S/4HANA - involve heavy upfront cost and long payback periods. When discount rates rise and uncertainty increases, boards delay irreversible commitments.

AI increases that uncertainty.

If automation changes workflows or operating structures, committing £50m–£100m to a three-year transformation can feel premature. Waiting carries option value.

The pattern that follows is rational.

Protect margins

Delay large Capex

Experiment with AI overlays

At the same time, shareholders expect productivity gains linked to AI. Workforce reductions are easier to execute than full operating model redesign. The risk, as HBR suggests, is sequencing: cutting cost before productivity mechanisms are embedded.

This tension is visible. Some large transformation programmes are postponed. Specialist SAP consultants report greater availability. System integrators face utilisation pressure. Not because ERP is obsolete, but because capital is cautious.

Yet AI strengthens the case for clean core systems. Advanced automation depends on structured data, defined ownership and coherent process architecture. Layering AI onto fragmented legacy estates may amplify complexity rather than reduce it.

This is not a story of collapse.

It is about timing.

AI accelerates expectations.

Higher rates slow capital commitment.

Organisations redesign more slowly than markets react.

For now, the firms that navigate this well may not be the most aggressive adopters. They may be the most disciplined in sequencing change.

In enterprise technology, boring remains resilient.