Cancer is a redox disease. Low levels of reactive oxygen species (ROS) are beneficial for cells and have anti-cancer effects. ROS are produced in the mitochondria during ATP production by oxidative phosphorylation (OXPHOS). In the present review, we describe ATP production in primary brain tumors, glioblastoma, in relation to ROS production. Differentiated glioblastoma cells mainly use glycolysis for ATP production (aerobic glycolysis) without ROS production, whereas glioblastoma stem cells (GSCs) in hypoxic periarteriolar niches use OXPHOS for ATP and ROS production, which is modest because of the hypoxia and quiescence of GSCs. In a significant proportion of glioblastoma, isocitrate dehydrogenase 1 (IDH1) is mutated, causing metabolic rewiring, and all cancer cells use OXPHOS for ATP and ROS production. Systemic therapeutic inhibition of glycolysis is not an option as clinical trials have shown ineffectiveness or unwanted side effects. We argue that systemic therapeutic inhibition of OXPHOS is not an option either because the anti-cancer effects of ROS production in healthy cells is inhibited as well. Therefore, we advocate to remove GSCs out of their hypoxic niches by the inhibition of their binding to niches to enable their differentiation and thus increase their sensitivity to radiotherapy and/or chemotherapy.