AI projects are multiplying. Budgets too. But measurable ROIs remain rare. Why?
Not because AI doesn't work. Because the conditions are not met. This is the observation we shared with SAP and Snowflake during SAP Connect on March 19.
And this is the starting point of a reflection we have been conducting for several years with our clients: Data & AI ROI is not achieved by deploying a model. It is built, step by step, on solid foundations.
The problem: SAP data trapped in its silos
In most organizations, ERP data is both the most valuable and the most difficult to activate.
Finance, customers, products, supply chain – everything is there, structured, reliable. But siloed.
As a result, when we want to cross-reference:
SAP data with IoT data, and,
Marketing data or external flows.
Projects run into costly ETL pipelines, a loss of business context, poor governance, and limited scalability. The cost of extraction destroys part of the value before AI has even started working.
It is precisely this problem that the partnership SAP × Snowflake addresses.
The two conditions for AI ROI
We now structure our response around two concepts that we systematically use in our missions:
AI Readiness – Is your data ready to feed AI?
This covers governance (certified, traceable, reliable data), building consumable data products, and semantic enrichment of the core. Without AI Readiness, AI trains on noise.
AI Gravity – Does AI run where your data is?
The principle is simple: moving data is expensive and creates risks. AI must run as close to the data as possible, not the other way around. This is what Cortex AI on Snowflake and Joule on SAP allow – a native execution, without duplication.
When one of the two conditions is missing, the AI initiative fails. No ROI. The equation we propose: AI Readiness + AI Gravity + Integrator expertise = Measurable Data & AI ROI.
A concrete case: €4.6M per year in heavy industry
To illustrate what these conditions make possible, we presented the case of a European manufacturer of heavy equipment whose ambition is to grow its aftermarket revenues from $120M to $180M by 2029.
-> The workflow is as follows:
connected equipment (IoT devices) send telemetry and health monitoring data in real-time.
A machine learning model calculates the time before a potential failure.
As soon as an alert is generated, the system cross-references it with SAP data – Is the contract active? Is there a maintenance service subscribed? If not, an alert is automatically sent to the salesperson in charge of the customer to offer proactive intervention.
-> This workflow: Semi-structured IoT data + structured ERP data + AI + business process – generates an estimated value of €4.6M per year across three use cases:
Operational maintenance,
On-site assistance,
Remote assistance.
What makes this case possible? The combination of SAP BDC (business context, governance) + Snowflake (integration of external IoT data, AI execution) + KPC (architecture, integration, trajectory).
The Snowflake Zero Copy connector: what changes structurally
The partnership SAP × Snowflake relies on a core technical mechanism: Zero Copy. SAP data is accessible in Snowflake without physical duplication. No ETL pipelines to maintain. The SAP business context is preserved. Governance is simplified. No additional silos are created.
This is a foundational shift. For 20 years, Data and Process have grown separately in organizations. AI is what unites them – provided you have the right foundation to do so.
The recommended trajectory: Open → Enrich → Activate
We do not believe in a big bang. What we recommend to our clients is a three-step trajectory:
Step 1 -> Open.
Make ERP data accessible via SAP BDC (zero-copy). Immediate quick win, without overhauling the existing system.Step 2 -> Enrich
Connect external data – IoT, marketing, logistics – on Snowflake.
The informational scope widens, AI models have more material.Step 3 -> Activate.
Launch the first cross-functional AI use cases and feed the results back into SAP. Processes become augmented.
At each step: governance, no silos, quick wins combined with building durable foundations.
The role of KPC
KPC acts as the third pillar of this equation. Our double expertise – SAP world and data & AI world – allows us to activate the two following conditions simultaneously:
Build AI Readiness (governance, data products, semantics) and deploy AI Gravity (Snowflake architecture, Cortex AI, SAP BDC integration).
Our concrete entry point:
an AI Readiness Assessment mission
evaluating the maturity of your data foundation for AI,
identifying quick wins, defining the trajectory.
