Hardware by far! It would pay itself off after 5-7 months (that's including the electric bill)
By my Calculations, even at a measly 5% difficulty increase per re-target, the hardware wouldn't have paid off even after a year.
And that's assuming you could get $5000 worth of Mini-rig Value hashing power delivered today.
From my calcs, you'd only have 821 BTC after a year, whereas you could buy 1000 BTC now and save yourself the electricity/heat/hassle of dealing with the hardware.
This calculation is assuming:
- $5000 worth of "MiniRig Value" hardware
- No Delivery Wait Time
- No Electricity cost
- Difficulty increase of 5% every 13 days
- Block split in 210 days
- 1000 BTC bought today at $5 ea
"Growth factor" should read "Daily reward multiplier"
EDIT: This is not an attack on people buying/making hardware or bitcoins. I'm looking for a logical discussion.
I am fully aware that a 5% difficulty increase cannot be/is impossible to guarantee/foresee, but it's the best estimate I can come up with.