Manufacturing companies operate in a complex and volatile economic environment where accurate financial forecasting is essential for strategic planning, production optimization, and risk management. Traditional forecasting methods often fail to capture the non-linear and dynamic relationships between financial and manufacturing indicators, limiting predictive accuracy. While machine learning (ML) and deep learning (DL) approaches have gained traction, there is a lack of comprehensive studies comparing a wide range of models using real-world financial data tailored to manufacturing needs. This study fills that gap by evaluating traditional statistical models (ARIMA, ARIMAX, SARIMAX), ML algorithms (Gaussian Process Regression, SVR, MLP, Random Forest, XGBoost), and DL architectures (LSTM, RNN, N-BEATS, TFT) using S&P 500 index data from 2010 to 2017, incorporating 82 financial variables relevant to the manufacturing sector. To our knowledge, this is the first large-scale comparison to include N-BEATS and TFT models in this domain. Model performance was assessed using RMSE, MAE, MAPE, and MSE, revealing that DL models—particularly N-BEATS and RNN—consistently outperformed others in both accuracy and stability. This study highlights the strategic value of AI-driven forecasting in capital investment, production scheduling, inventory control, and financial risk mitigation. Future work should focus on real-time forecasting and the integration of external economic indicators to enhance financial decision-making in manufacturing.