Little is known about how alcohol consumption promotes the onset of human breast cancer(s). One hypothesis is that ethanol induces metabolic changes in the tumor microenvironment, which then enhances epithelial tumor growth. To experimentally test this hypothesis, we used a co-culture system consisting of human breast cancer cells (MCF7) and hTERT-immortalized fibroblasts. Here, we show that ethanol treatment (100 mM) promotes ROS production and oxidative stress in cancer-associated fibroblasts, which is sufficient to induce myofibroblastic differentiation. Oxidative stress in stromal fibroblasts also results in the onset of autophagy/mitophagy, driving the induction of ketone body production in the tumor microenvironment. Interestingly, ethanol has just the opposite effect in epithelial cancer cells, where it confers autophagy resistance, elevates mitochondrial biogenesis and induces key enzymes associated with ketone re-utilization (ACAT1/OXCT1). During co-culture, ethanol treatment also converts MCF7 cells from an ER(+) to an ER(-) status, which is thought to be associated with "stemness," more aggressive behavior and a worse prognosis. Thus, ethanol treatment induces ketone production in cancer-associated fibroblasts and ketone re-utilization in epithelial cancer cells, fueling tumor cell growth via oxidative mitochondrial metabolism (OXPHOS). This "two-compartment" metabolic model is consistent with previous historical observations that ethanol is first converted to acetaldehyde (which induces oxidative stress) and then ultimately to acetyl-CoA (a high-energy mitochondrial fuel), or can be used to synthesize ketone bodies. As such, our results provide a novel mechanism by which alcohol consumption could metabolically convert "low-risk" breast cancer patients to "high-risk" status, explaining tumor recurrence or disease progression. Hence, our findings have clear implications for both breast cancer prevention and therapy. Remarkably, our results also show that antioxidants [such as N-acetyl cysteine (NAC)] can effectively reverse or prevent ethanol-induced oxidative stress in cancer-associated fibroblasts, suggesting a novel strategy for cancer prevention. We also show that caveolin-1 and MCT4 protein expression can be effectively used as new biomarkers to monitor oxidative stress induced by ethanol.