A longstanding criticism to Keynesian and Kaleckian growth theories is the question: why would firms operating with underutilized capacity still accumulate capital stock? This paper offers an answer by analyzing the choice of capacity utilization and accumulation in a strategic setting. The argument hinges on the Keynesian notion of user cost of capital. We argue that firms have incentives to wait to see what other firms are doing before adjusting their own utilization, which we capture through a marginal user cost of own utilization decreasing in average utilization. Accordingly, interactions among firms involve strategic complementarities: it is profit-maximizing to increase own utilization with average utilization. Since the latter is a reasonable proxy for demand, (i) the analysis provides a rationale for treating desired utilization as endogenous to demand at the firm level. In general equilibrium: (ii) capital accumulation coexists with underutilization; (iii) if firms were able to coordinate on a common utilization rate, utilization would be strictly higher than in equilibrium. The implications for growth and distribution depend on how the model is closed: (iv) with a distributive closure, equilibrium growth and profitability are both strictly below their socially-coordinated counterpart; (v) with an exogenous labor supply closure, the equilibrium labor share is strictly smaller than under coordination. Hence, (vi) there are mutually beneficial bargaining opportunities for both capital and labor. Moreover, (vii) demand policies have multiplier effects. The slow recovery from the Great Recession in the US provides a prime example of the relevance of equilibrium underutilization. Finally, we use stateby-sector data from the BEA to validate our hypothesis: (viii) our estimation results provide strong and robust support for the relevance of strategic complementarities in the US.