Artificial intelligence (AI) has shifted from being a promising accelerator to a force that is reshaping long-standing assumptions about enterprise software architecture. Nowhere is this change more visible than in ERP.
When Anthropic released its Financial Analysis Solution earlier this year, the product generated meaningful ERP-style financial analytics.
Such offerings do not require customers to complete lengthy upgrades or restructure their environment into a clean core; they provide value immediately through APIs and code, customisation and process management.
Around the same time, hyperscalers introduced AI services that connect directly into legacy SAP estates and automated reasoning engines. Together, these developments point to a profound change in how ERP capability is created, consumed and valued as the stack that once moved only in gradual increments is beginning to transform in unexpected ways.
The pressure AI is placing on ERP
ERP systems have historically depended on well-defined layers of technology. Databases anchor the bottom of the stack, business logic and process controls sit above them, and user interfaces and reporting tools occupy the top. This architecture has served large enterprises for decades because it produces reliability and predictability.
AI works differently. Foundation models interpret context from wherever it exists. They can extract meaning directly from data structures, generate workflows dynamically, and interpret complex business states without relying on pre-built process logic. Instead of integrating into the stack at a single point, they move across it and interact with multiple layers at once.
The outcome is a new pressure on the ERP market. Vendors must now demonstrate the enduring value of their platforms as AI becomes capable of synthesising certain layers of functionality on its own.
The remaining differentiators for ERP providers lie in areas that require strong discipline and governance. Large enterprises depend on systems that deliver reliable transactional integrity in order to meet regulatory and internal audit requirements, and to maintain long-term data consistency across thousands of interconnected processes and applications. These are the domains where ERP will continue to matter, even as AI surrounds and speeds up everything else.
SAP’s clean core strategy and why it may help enterprises rather than restrict them
SAP continues to play a pivotal role in enabling customers to fully leverage its embedded AI capabilities. Running modern versions of S/4HANA, adopting clean core principles, and tidying up decades of customisation are all part of that equation. For organisations balancing day-to-day operations with large-scale transformation, these requirements can feel demanding. However, when viewed through the lens of AI readiness, they also provide a clearer framework for how value can be realised over time.
AI amplifies the quality of the data it consumes. When processes are inconsistent and when data structures vary across business units, AI produces unreliable outcomes. The organisations that are currently seeing early value from SAP’s AI tools are those that completed their clean core journeys early. Now that their data is coherent, processes are stable and technical debt is manageable, they are in a stronger position to take advantage of SAP’s roadmap as it matures.
For many businesses, the clean core journey is beginning to look less like a vendor requirement and more like a strategic path toward scalable AI adoption within ERP. At the same time, customers are recognising that value can be unlocked in multiple ways and at different speeds. Understanding how and when to align with SAP’s approach, while also making pragmatic use of external capabilities, is becoming a critical part of the AI conversation.
Hyperscalers are creating a parallel AI path for legacy ERP estates
Alongside SAP’s longer-term AI strategy, hyperscalers are addressing a more immediate customer need. They are targeting the large segment of the market that still runs older versions of SAP or heavily customised estates where upgrades are complex and costly. These customers want the benefits of AI but cannot simply restructure their business processes overnight.
Hyperscalers have begun delivering AI capabilities that work directly with legacy ERP environments. Traditionally, any mismatch between an invoice and a purchase order triggered manual intervention and enterprises outsourced these processes to global BPO providers. Now, new AI tools from hyperscalers can already handle the data retrieval, vendor communication and contextual reasoning required to propose resolutions. In some tests, this approach has reduced BPO costs for targeted use cases by as much as 75 percent.
If these numbers hold at scale, they will reshape the economics of BPO and place new competitive pressure on ERP vendors. Customers may delay migrations if AI allows them to extract value from their existing environments. ERP providers will need to demonstrate that their AI-enabled future offers richer and deeper benefits than the bolt-on tools offered by hyperscalers.
Governance challenges are rising as quickly as the opportunities
The enthusiasm surrounding AI often overlooks an emerging concern. As enterprises increasingly experience AI-driven security incidents, attackers are learning to persuade AI interfaces to bypass safety measures and reveal information or perform actions that users did not intend.
These developments are forcing organisations to confront the immaturity of their AI governance frameworks. Cost metering, security operations for AI agents and clearer enterprise control mechanisms will all rise in importance in 2026 as the industry moves out of its early, experimental phase. What looked adequate a year ago is already struggling to keep up with the pace of innovation.
A more competitive landscape than ERP has seen in years
The year ahead will be transformative for ERP. Vendors will need to articulate the durable value of their platforms in a world where AI can generate workflows and logic rapidly. At the same time, hyperscalers will keep expanding AI tools that reduce the friction of older ERP systems, while enterprises will face urgency to modernise their data and process foundations so that AI becomes a complementary force rather than becoming a new source of risk.
ERP will not disappear because its role as the source of truth for global organisations remains essential. But what is changing is the structure of the ecosystem around it. AI is no longer an add-on to ERP; it is increasingly becoming a parallel engine that can replace, enhance or compete with significant parts of the stack.
2026 will reward the organisations that prepare for this shift early. It will be far less forgiving to those that assume the old architecture will hold.
