Check http://www.ncbi.nlm.nih.gov/pubmed/19572193
Eyeballing their figure seems to indicate that the g factor is much less strong in these children. A quick search on Scholar didn’t reveal any studies that investigated this idea.
If someone can obtain subtest data from autism samples, that would be useful. The methods I used in my recent paper (section 12) can estimate the strength of the general factor in a sample. If g is weaker in autistic samples, this should be reflected in these measures.
I will write to some authors to see if they will let me how the subtest data.