SAP's Chief Quantum Officer: AI Is Commoditizing Prediction, So Decision-Making Is Now the Edge

Artificial intelligence is fast becoming table stakes — within a few years, nearly every large company will have roughly the same level of forecasting capability. Carsten Polenz, SAP's Chief Quantum Officer, argues that once prediction becomes a commodity, it stops being a source of competitive advantage. The next frontier, he says, isn't predicting what will happen — it's deciding what an enterprise should do amid thousands of interconnected choices, conflicting goals, and limited resources.

Polenz illustrates this with a simple example: at the end of every quarter, a payments team holds back payments to preserve liquidity, a receivables team accelerates collections to hit its targets, and a sales team decides which deals to push, which disputes to escalate, and which customer gets a discount. Each department makes a locally sensible call, yet the aggregate outcome isn't the one the enterprise as a whole would have chosen. AI can predict which deal is likely to close or flag which receivable is at risk, but it doesn't answer the real question: what should the company actually do?

To fill that gap, Polenz proposes a new corporate technology category he calls Enterprise Decision Computing. The approach treats a business decision — its possible actions, goals, constraints, uncertainty, interdependencies, and economic consequences — as a single corporate object that can be modeled and optimized holistically. Resource-planning systems execute processes, business intelligence explains the past, and AI forecasts outcomes — but none of these tools, alone or combined, answer the question of what coordinated set of actions an enterprise should take given its goals, constraints, and the dependencies between its departments.

The debate around quantum computing, Polenz argues, is currently pointed in the wrong direction — all the attention goes to hardware milestones like qubit quality and error correction, when the real question isn't when quantum technology will be ready, but what it should actually compute once it is. His approach follows a principle of progressive enrichment: start with a classical model factoring in revenue, deal-closing probability, and sales resources, then add margin and payment terms, then cash flow and supply constraints, and finally portfolio-level interactions. Each additional layer makes the decision more realistic — but also harder to compute — and it's precisely at the point where that complexity overwhelms classical methods that quantum computing finds its role.

Polenz offers executives a concrete recommendation: identify one high-frequency, high-impact area — sales, payables, or receivables, for instance — that departments are currently optimizing independently, stand up a baseline model, and compare the coordinated answer against what each department would have decided on its own. "Decision debt," he says, accumulates quietly, much like financial debt, until it becomes a serious problem. The investment required is relatively modest — but the cost of not having this insight, while competitors are already acting on it, can be far higher.