A two-period symmetric Cournot duopoly with linear demand and costs is analyzed under linear (or more general) returns to scale in process R&D. Subgame-perfect equilibrium may call for one firm to fully innovate while the other firm remains just as before. The outcome is a polar duopoly or monopoly (one firm endogenously exiting). Two research joint venture schemes and the noncooperative solution are compared. Due to built-in symmetry, a joint lab does not always lead to the best performance. Overall, our findings differ quite substantially from those based on strongly decreasing R&D returns and symmetric outcomes