Cost is important to planning manufacturing, cost creates a direct effect on profit and price setting over the unstable supply chain. Neural Network (NN) is one widely used artificial intelligence technique for prediction and recognition in various applications that can be applied for estimating the cost of manufacturing, although requiring a complex multi-layer structure and not specifying to the cost estimation purpose. This paper presents the ongoing research for creating the Time-Driven Cost Estimation Model (TDCE), which is an artificial neural network-inspired architecture that uses a well-known cost estimation technique Time-Driven Activity-based Costing (TDABC) as the driver of a model. The first step of the methodology starts with modeling the equation of the overall TDABC resource group to suit the neural network structure, then designing the architecture that one resource group model can consist of four resource group elements – materials, daily paid labor, monthly paid labor, and capital cost. We establish the model with hidden weights across the resource group and elements level. The gradient descent is applied as a weights learning technique to be adjusted in each iteration. We simulate the dataset and train the resource group model. The result shows that the purpose model can produce accurate results for the non-complicated dataset.