Productive energy (PE) predicts performance better than apparent metabolizable (AME), N-corrected AME (AMEn), and classic net energy (CNE; CNE = AME – heat increment, HI). This study assessed the relationship between body composition, fasting heat production (FHP = net energy for maintenance, –NEm), and economics, and explored why PE outstands. 1920 chicks in 96 pens (8 blocks) were assigned to 12 dietary treatments with changing levels of total digestible amino acids (TDAA; T1-T3), digestible starch (sSt) coupled with TDAA (T4-T6), nutrient density (T7-T9: fat, TDAA; T10-T12, fiber). During one week, each block was restrict-fed treatment diets allowing similar energy intake; every block started the experimental period every seven days. Before that, all received standard diets. For each block, performance was recorded, body protein, fat, and energy (net energy for gain, NEg) and processing weights (i.e., breast meat, leg quarters; breast-to-leg ratio, BLR) were determined with dual-energy X-ray absorptiometry. Heat production (HP) at fasting (FHP) and HI (HI = fed HP – FHP) were determined in calorimetry chambers, and PE (NEg+NEm) was calculated. Diet AMEn, non-starch polysaccharides fractions, and digestible nutrients (fat, crude protein –CP, starch, amino acids) were determined. Carcass market value (MKV) was calculated with market prices. Gross profit gain (GPG) was determined as MKV – feeding cost. Linear mixed models were fitted in JMP to quantify empirical relationships, and a mechanistic model was developed. Models indicated positive linear relationships (P<0.001) between dCP intake and NEm (kcal/bird/d; adj.R² = 0.98), body gain leanness with NEm (adj.R² = 0.96) and BLR (adj.R² = 0.94), BLR and MKV (adj.R² = 0.93), and MKV with GPG (adj.R² = 0.99). The models indicated that in contrast with PE, AME (and AMEn and CNE) (i) do not acknowledge (P>0.05) changes in body composition, their influence on FHP and BLR, and its effect on MKV and GPG, (ii) mislead the influence of CP on energetics by ignoring (P>0.05) its effect on NEm, and (iii) assumes no variations in FCR (P>0.43; other than digestion) and actual metabolism.