For years, executive discussions about enterprise systems have followed a familiar script. Someone raises the need to modernise finance or supply chain and the conversation quickly narrows to which large ERP suite to standardise on, how many years the programme will take and how much organisational pain everyone is prepared to tolerate.
Lately, a different kind of question has started to appear in those rooms. It is no longer just “Which ERP do we choose?” It is: what happens when AI stops being a bolt-on feature and starts orchestrating the way the enterprise actually runs?
That shift sounds abstract, but its effects are practical. When decisions, workflows and controls live in intelligent layers that span multiple systems, the ERP does not disappear. It becomes something quieter and less visible: a ledger of record, a backbone for controls, an essential but largely invisible part of the plumbing while the centre of gravity moves elsewhere.
Understanding this change is now one of the most important strategic tasks for senior leaders.
ERP was built for a world where systems were the destination
Traditional ERP earned its place at the centre of large organisations for good reasons. It offered a single, structured environment in which critical processes could be standardised. Auditors knew how to interrogate it, regulators understood its control frameworks and boards could approve eye-watering budgets on the basis that they were buying predictability, traceability and long-term stability.
In that world, it made sense to put as much work as possible into ERP. The system was where process logic lived, where users logged in each day, where transactions were keyed and approved. Choosing a platform was almost synonymous with choosing an operating model.
The world now forming around AI does not work that way. The system of record still matters, but it is no longer the primary place where people think and decide.
AI is turning ERP into infrastructure
As AI capabilities mature, the most interesting use cases are not smarter reports but new patterns of work, where month‑end activities run continuously instead of in a frantic burst, exceptions are identified and resolved before anyone notices, and supply and demand signals are quietly rebalanced long before they make it onto a slide deck.
In these scenarios, the interface is not a transaction screen. It is an assistant, an agent or a conversational front end that draws on data from multiple sources at once. Users ask questions, set objectives and review exceptions. The underlying systems, including ERP, are there to provide truth, structure and control, but they are not where most people spend their day.
Once that happens at scale, a subtle inversion occurs. ERP remains vital, especially for heavily regulated, high-risk processes, but it behaves more like a utility service: always on, rigorously governed, rarely glamorous. The strategic contest moves to the orchestration layer above it.
A wider set of options than most boards acknowledge
While this shift has been building, a broader ecosystem of alternatives has matured. There are cloud-native platforms that began with finance and have expanded into adjacent domains. There are open-source systems that now offer full ERP capabilities, and there are specialist and industry-specific platforms that handle particular slices of the value chain exceptionally well. Many of them embed AI into their core from the outset.
These tools are no longer confined to experimental pilots. They are running real revenue, inventory and operations in organisations that are not small. Often, they appear first in new subsidiaries, spin-offs or digital ventures where speed, flexibility and cost discipline matter more than following historic patterns.
The lesson isn’t that traditional ERP is obsolete, but that enterprises now have far more credible options than many currently act as if they do.
Thinning the core, not ripping it out
Where a thick, traditional core is not strictly required, the calculation looks very different. Business units with relatively simple requirements, internal reporting domains and new ventures that need to be live in months rather than years, experience a different kind of pressure. In those spaces, the cost and rigidity of applying the same heavy solution everywhere becomes harder to defend.
That is where newer platforms and AI-driven orchestration are already taking hold. They run real workloads at lower cost, on cleaner data models, with shorter implementation cycles and automation built in rather than bolted on. Crucially, they do this alongside the traditional core rather than trying to displace it on day one.
As that coexistence matures, the landscape changes. More process logic and day-to-day decision-making move into AI layers that sit across systems. The old ERP estate retreats to the places where intense scrutiny and formal assurance are non-negotiable. Everywhere else, work is mediated and optimised elsewhere. At that point, the question for leaders is no longer “Could we ever replace this system?” but “Exactly what job are we still asking it to do, and is there a leaner way to provide that?” ERP strategy becomes less a single vendor choice and more an exercise in deciding where different levels of control genuinely belong.
Some activities will always justify dense, proven, heavily audited infrastructure. Others can be safely governed through AI-rich orchestration sitting over a simpler ledger. Still others, especially at the edge of the organisation, may be better served by platforms that make different assumptions about architecture and economics. The hard work is not the software selection itself, but the honesty required to admit which processes truly demand maximum conservatism and which have been dragged into that category by habit.
Closer than a distant hypothesis
None of this means large enterprises are on the verge of handing their most critical finance processes to unproven platforms and opaque models. At the core, a degree of caution remains essential, and in that sense the world is not yet ready for a wholesale replacement of traditional ERP.
It would, however, be a mistake to treat this as a remote, theoretical horizon. AI is already changing how people interact with core systems. Alternative platforms are already carrying meaningful slices of revenue and operations. The pressure to separate the areas that truly need heavyweight ERP from those where it has simply become the default is only moving in one direction.
The limiting factor now is less the capability of the technology and more the organisation’s ability to live with a hybrid core: a deliberately thinned centre of control, surrounded by faster moving, AI-orchestrated layers. Those who continue to treat ERP as a solved, static problem may discover, sooner than they expect, that the real contest for value and differentiation has shifted to that orchestration layer, and that competitors have been redesigning around that reality while they stood still.
