Abstract: In this paper, the cell construct in biology, the process construct in operating system, and the grip construct in grids are compared. It is noted that the process construct in traditional operating systems is very similar to a bacteria cell. To accommodate the complexity caused by dynamic and open characteristics of grids, grid needs to learn from more advanced life forms, such as plants and animals, which provide for proliferation, differentiation, and selection. This leads to two new models of computing, called development computing and natural computing, that could facilitate autonomous computing in a grid. The state transitions of cells are analyzed from an operating system’s viewpoint, the differences are pointed out, and essential features needed by development computing are identified. Based on this analysis, a grid process construct that supports development computing is proposed.