Artificial Intelligence (AI) is reopening a strategic debate many enterprises believed had largely been settled over the past decade: should companies build their own technology or rely on external platforms?
In procurement, this question is becoming increasingly relevant for Source-to-Pay (S2P) technology.
- A familiar debate in enterprise technology
For many years, the dominant approach was clear: organizations adopted specialized SaaS procurement platforms and configured them to their operating model. Solutions such as Coupa , Ivalua , SAP Ariba , Oracle, GEP Worldwide , penny. or Zycus enabled companies to standardize processes, accelerate deployment, and benefit from continuous innovation delivered by the vendor ecosystem.
But AI is changing the equation.
Many organizations are now reconsidering whether they should develop their own AI capabilities for procurement, rather than relying exclusively on the AI features embedded within external S2P platforms.
Interestingly, this debate strongly echoes the evolution of enterprise technology over the past three decades.
During the ERP era, companies invested heavily in customizing systems to fit their specific processes. While this created powerful capabilities, it also introduced significant complexity and technical debt. The shift to SaaS in the 2010s encouraged a different mindset: standardize processes, configure rather than customize, and benefit from vendor-driven innovation.
Today, AI is bringing us back to a strategic crossroads.
- The case for building AI internally
Building internal AI capabilities can unlock significant strategic value for procurement organizations.
Procurement functions hold large volumes of highly valuable data: supplier performance history, negotiation outcomes, contract obligations, pricing benchmarks, and risk indicators accumulated over many years.
When leveraged correctly, this data can power advanced AI use cases such as:
- supplier risk prediction
- automated spend intelligence
- negotiation insights
- contract analytics
- Procurement copilots supporting category managers
Developing these capabilities internally allows organizations to create proprietary intelligence that competitors cannot easily replicate. It also enables companies to tailor AI to their specific procurement operating model, governance structure, and category strategies.
However, building AI internally comes with significant challenges.
Successful AI development requires strong data foundations, robust data engineering, machine learning expertise, and governance frameworks to manage model performance, security, and compliance.
For many procurement organizations, the primary obstacle is not the AI itself, but the data landscape. Procurement data is often fragmented across ERP systems, S2P platforms, supplier portals, and external risk databases. Without a clean and integrated data foundation, AI initiatives quickly lose reliability and credibility.
There is also the question of cost and sustainability. Developing and maintaining AI capabilities requires ongoing investment in infrastructure, engineering talent, and continuous model improvement.
- The value of vendor-driven AI
At the same time, procurement technology providers are investing heavily in AI capabilities embedded directly into their platforms.
Leading S2P solutions are introducing features such as autonomous sourcing, supplier discovery, contract intelligence, and procurement copilots integrated into daily workflows.
Because these vendors operate at scale, serving many organizations across industries, they can continuously improve their models and distribute innovation through product updates.
However, vendor AI also has limitations.
By necessity, these models remain relatively generic. They must serve a broad customer base and cannot fully capture the unique supplier ecosystems, negotiation dynamics, or category strategies of individual organizations.
- The emerging hybrid model
As a result, many organizations are now moving toward a hybrid architecture for procurement AI.
In this model, the S2P platform remains the transactional backbone, managing sourcing events, purchasing workflows, invoicing processes, and supplier management.
On top of this foundation, companies build an internal AI intelligence layer using enterprise data platforms. This layer enables proprietary analytics, decision support, and automation capabilities tailored to the organization’s procurement strategy.
This approach mirrors how enterprises evolved beyond ERP systems. The ERP remained the system of record, while innovation and intelligence moved into data platforms and analytics ecosystems built around it.
The same pattern is now emerging in procurement.
The S2P platform manages transactions, while procurement intelligence increasingly resides in enterprise data and AI layers controlled by the organization.
- A strategic question for procurement leaders
Ultimately, the debate is no longer simply build versus buy.
The real strategic question is:
Which parts of procurement intelligence should remain proprietary?
