No hindsight allowed: the UPC rejects inventive step attacks that are based on fragmented analyses (UPC_CFI_337/2025)
UPC Case Law | 24.04.2026
Court docket: CD Munich, Decision of 24.02.2026
UPC_CFI_337/2025 [EP 3 296 274]
Parties: TCL EUROPE SAS v. Corning Incorporated
Contributor: Marie Isabel Thomson
Headnote
1. A realistic starting point is typically a prior art disclosure as a whole. Absent a specific reason or pointer in the disclosure itself (or based on common general knowledge) to do so, the selection of a particular example composition as a “starting point”, merely because it happens to come “closest” to the claimed subject matter in terms of structural components, bears the risk that such selection itself already involves hindsight.
2. Where the features of a patent claim, in an interdependent way, even if they are not synergetic in the sense of having a special combination effect, provide a solution to the objective problem, ignoring these interdependencies and dividing the objective problem up into separate problems amounts to hindsight reasoning which is to be avoided in the assessment of inventive step.
Relevance of the decision
The present UPC Order UPC 337/2025 of the Munich Central Division provides important clarification on the methodology for assessing inventive step.
Specifically, the Court emphasised that a prior art document must be assessed as a whole. In doing so, it cautioned against selecting a specific embodiment or example from within a broader disclosure as the starting point for the analysis, unless the prior art itself provides a reason or pointer for doing so. In the absence of such a pointer, isolating a particular embodiment or example, bears the risk that such a selection itself already involves hindsight, as the selection may already be influenced by knowledge of the invention. The Court thus underscored that the skilled person would not arbitrarily focus on a specific example, but would instead derive their starting point from the overall technical teaching of the document.
Thus, while EPO practice often starts with a specific embodiment from within the closest prior art based on the criterion of “most features in common”, the UPC decision highlights that such a selection must itself be free from hindsight.
The Court furthermore held that it is not permissible to artificially split the claimed subject-matter into individual features and assess their obviousness in isolation. The skilled person interprets individual claim features in light of the claim as a whole and in view of their technical function within the overall teaching (the inventive concept). In the present case, the Court concluded that the skilled person, reading the claim as a whole in the context of the description, using its common general knowledge, would understand that the claim features are interdependent. Against this background, the inventive step assessment must consider whether the skilled person would have been led to the claimed combination as a whole, rather than treating the features as independent building blocks that can be evaluated separately.
The Court thus rejects a fragmented, feature-by-feature analysis, if the claim features are interdependent even without synergy. By contrast, the EPO often requires the presence of a synergistic effect to support an inventive step based on a combination of features (CLB I.D.9.3.2).
Overall, the UPC Order appears more patentee-friendly than the general approach taken at the EPO on these points.