A large number of genomic studies have reported associations between the gut microbiota composition and metabolic diseases such as obesity or type 2 diabetes. This led to the widespread idea that a causal relationship could exist between intestinal microbiota and metabolic diseases. At odds with this idea, some compelling studies reported that global changes in microbiota composition have no effect on the host metabolism in obese mice or humans. However, specific bacteria are able to confer host metabolic benefits, such as Akkermansia muciniphila or Prevotella copri, when they are given by gavage in obese mice. A crucial link by which gut bacteria communicate with the host mucosa is based on metabolites or low-molecular-weight compounds. Among them, short-chain fatty acids produced from the fermentation of dietary fibers initiate beneficial effects on the host metabolism via the activation of intestinal gluconeogenesis (a mucosal function exerting antidiabetic and antiobesity effects through the activation of gut-brain neural circuits). However, fermentation of short-chain fatty acids is a function that is widespread among the main bacterial phyla and thus weakly depends on microbiota composition. Therefore, even if some bacteria may confer on the host metabolic benefits, the causal role of microbiota in metabolic diseases is not established.