Here are samples of data from my 13 inch MacBook Pro with 8 GB of RAM before and after opening Chrome:
Mac:~ bmike$ vm_stat
Mach Virtual Memory Statistics: (page size of 4096 bytes)
Pages free: 212512.
Pages active: 931907.
Pages inactive: 381136.
Pages speculative: 55403.
Pages wired down: 515533.
"Translation faults": 521414918.
Pages copy-on-write: 9952178.
Pages zero filled: 465150147.
Pages reactivated: 93725.
Pageins: 1567093.
Pageouts: 6081.
Object cache: 100 hits of 17386036 lookups (0% hit rate)
Mac:~ bmike$ open -a "Google Chrome"
Mac:~ bmike$ vm_stat
Mach Virtual Memory Statistics: (page size of 4096 bytes)
Pages free: 136402.
Pages active: 990000.
Pages inactive: 392735.
Pages speculative: 53707.
Pages wired down: 523945.
"Translation faults": 521638787.
Pages copy-on-write: 9972877.
Pages zero filled: 465252021.
Pages reactivated: 93725.
Pageins: 1577678.
Pageouts: 6081.
Object cache: 100 hits of 17386267 lookups (0% hit rate)
The setup of the test is a system that's been running for 6 days, and is otherwise not misbehaving. Here are the commands I run to make sure the system is in balance:
Mac:~ bmike$ iostat 5
disk0 cpu load average
KB/t tps MB/s us sy id 1m 5m 15m
28.30 9 0.26 2 1 97 0.78 0.61 0.61
0.00 0 0.00 0 1 99 0.72 0.60 0.60
32.00 0 0.01 1 1 99 0.66 0.59 0.60
0.00 0 0.00 0 1 99 0.61 0.58 0.59
^C
Mac:~ bmike$ vm_stat 5
Mach Virtual Memory Statistics: (page size of 4096 bytes, cache hits 0%)
free active spec inactive wire faults copy 0fill reactive pageins pageout
210871 940704 53319 379363 512550 522503K 9987967 469254K 93725 1579260 6081
210776 931311 53771 379369 521225 5150 10 32953 0 0 0
211289 930966 53798 379369 521207 4606 0 57255 0 0 0
211321 930971 53799 379369 521207 4589 3 45933 0 0 0
211531 930998 53799 379369 520925 10999 0 77822 0 0 0
I set Google chrome up to have 10 blank tabs and quit it and then ran the test above to measure exactly how much RAM changes before and after launching it. As you can see, launching Chrome took about 76k free memory pages out for the free pool and caused the inactive pool to grow by about 11.5k pages.
Given these numbers, Chrome seems highly unlikely to be causing a leak itself and your web pages might be causing the load. If you can run these tools in terminal, you'll likely see that your system had other issues with memory management and Chrome is likely the straw that puts the final load on the system before it starts swapping. Look for paging activity before you launch Chrome and also have a look at my answer here on why inactive memory is almost never the actual problem and this post hopefully explains how to convince yourself whether you have an actual problem with the memory manager or a specific program.
Best Answer
Yes, that amount is normal.
As kernel_task performs a lot of tasks on behalf of user-space processes (i.e. the ordinary programs you're running on your computer), the actual memory usage of kernel_task will vary a lot according to which programs you're currently running, and what you're doing with them.