According to
a mining calculator, at current difficulty you would mining a block roughly every 47 days. At this rate, variance would still be very significant and you wouldn't want to live in the knowledge that you could have to keep living on the same block for a much longer time than expected at any point. For that reason, I would suggest using a pool.
But since pools can take fees and even if they're low they can be annoying, plus the Bitcoin network needs some help decentralising into a collection of many different pools, the best solution shouldn't be to
join a pool but to
start a pool. With your hashrate you would already be bigger than some smaller pools yourself, so I would start your own pool with negligible fees (maybe 0.1-0.3%) and then advertise your pool for its low fees and soon enough miners will join to decrease the variance. You should also limit your pool so that it couldn't have more than 1% of the network's total hash rate, in the spirit of decentralisation.
Even though with this setup you'd be generating a Bitcoin every ~4 days, you should still have a backup - at least have some skills that could be applicable to other jobs if the difficulty or state of the Bitcoin network changes and it becomes harder for you to earn.
Interesting replies from everybody so far! I hear you on the having a backup plan. I suppose an advantage of using S9s or any Sha256 compatible miner is that, if Bitcoin fails, there are other coins that are profitable today with the same equipment. Alternatively, you could liquidate your miners and sell them for cash for another venture! What are your thoughts there?
It seems like the rate of income from a total hashrate like 500TH/s or 1PH/s would pay the equipment off in 1-2 years, so as long as that goes well -- you're profiting. It doesn't seem any less risky to me than any other venture. I'd be willing to bet bars and restaurants fail with a much higher rate than cryptocurrency farms, to date! Though, I guess there's no way to prove that.