Can giant planets form by direct gravitational instability?

Gravitational instability has been invoked as a possible mechanism of giant planet formation in protoplanetary disks. Here we critically revise its viability by noting that for the direct production of giant planets it is not enough for protoplanetary disks to be gravitationally unstable. They must also be able to cool efficiently (on a timescale comparable to the local disk orbital period) to allow the formation of bound clumps by fragmentation. Combination of dynamical and thermal constraints puts very stringent lower limits on the surface density and temperature of disks capable of fragmenting into self-gravitating objects: for the gravitational instability to form giant planets at 10 AU gas temperature at this location must exceed 10^3 K for a minimum disk mass of 0.7 M_Sun and minimum disk luminosity of 40 L_Sun. Although these requirements relax in more distant parts of the disk, masses of bound objects formed as a result of instability even at 100 AU are too large (\sim 10 M_Jupiter) to explain characteristics of known extrasolar giant planets. Such protoplanetary disks (and planets formed in them) have very unusual observational properties and this severely constrains the possibility of giant planet formation by direct gravitational instability.