An increase in the riskiness of a technology will raise economy-wide expected output: large productivity realizations lead to gains while small realizations are mitigated by using alternative technologies. Some usage of even the riskiest technologies can therefore bring Pareto improvements. The observed expected output of risky technologies can nevertheless be less than that of safer technologies: empirical estimates of expected output are therefore a poor measure of efficiency. Firms will adopt the efficient riskier technologies when markets are competitive, contrary to the view that innovation requires a monopoly reward. These results require a suitable definition of risk for general equilibrium models and a refinement of competitive equilibrium that allows for information discovery. The efficiency gain of increases in risk is related to real options theories of investment but applies even to single-period economies.