Applying constant elasticity of substitution (CES) functions in general equilibrium integrated assessment models (GE-IAMs) for the substitution of technical factor inputs (e.g., replacing fossil fuels) fails to match historically observed patterns in energy transition dynamics. This method of substitution is also very sensitive to the structure of CES implementation (nesting) and parameter choice. The resulting methodology-related artifacts are (i) the extension of the status quo technology shares for future energy supply relying on fossil fuels with carbon capture, biomass, and nuclear; (ii) monotonically increasing marginal abatement costs of carbon; and (iii) substitution of energy with non-physical inputs (e.g., knowledge and capital) without conclusive evidence that this is possible to the extent modeled. We demonstrate these issues using simple examples and analyze how they are relevant in the case of four major CES-based GE-IAMs. To address this, we propose alternative formulations either by opting for carefully applied perfect substitution for alternative energy options or by introducing dynamically variable elasticity of substitution as a potential intermediate solution. Nevertheless, complementing the economic analysis with physical modeling accounting for storage and resource availability at a high resolution spatially and temporally would be preferable.