So right now (well one week ago when I originally wrote this) I can buy a 2 year 100 MH/s (mega-hashes per second) Ethereum mining contract from Genesis the cloud mining company for $2800 . According to various profitability calculators that would currently yield around 1.5 ETH per month  which is 36 ETH over 2 years. However if today I took that $2800 I could only buy 7.8 ETH at $360.
WOW! Holy smokes! What a deal!
The Gut Check
First gut check... if it is so profitable why isn't everyone doing this?
Second gut check... if it is so profitable why doesn't Genesis just use the mining hardware themselves and make a fortune?
If you're an optimist you might say "well, maybe people just don't know, just like they don't know what crazy returns ETH is giving on the USD market". Or they could say "maybe Genesis doesn't want to gamble that ETH will stay high - this way if ETH goes back to $2 they will still have made a bundle selling hardware?"
The Devil is in the Details
The reality is that the devil is in the details. Take a look at this chart: https://etherscan.io/chart/difficulty 
That tells you the ratio of hashes per block of ETH mined. Note how since Jan 1st 2017 to now it has gone up from about 80 TH to almost 800 TH (terra hashes). So the amount of ETH per MH/s your contract gets you has gone down by 10x in that regard. I've seen some data that say difficulty is currently increasing around 2.2% per day  so at the end of a 30-day period your yield per day is down to around 50% of what it was before and your total for the month is about 60% of what it would be if yield stayed constant.
Another way to look at the increasing difficulty is that you could say, well it just means it takes 10x longer to compute the hashes compare to 6 months ago but you still get ETH at the end, it just burns a bit more electricity getting there (well 10x is quite a bit more...) Remember also that at least with cloud mining the contract is limited to 2 years - for example if your first 1.5 ETH takes a month to mine, second 1.5 ETH takes 2 months to mine, and your third 1.5 ETH takes 4 months to mine then you're never going to see anything like 36 ETH at the end of 2 years, you just run out of time.
Also take a look at this chart: https://etherscan.io/chart/hashrate . That tells us that hashing has gone up from 5,700GH/s around Jan 1 to 48,000 GH/s which is an almost 10x increase. So now you have 1/10th the chance of discovering the next block compared to just over 6 months ago. Extrapolate this kind of exponential growth forward over 2 years and you should see that the yield is looking nothing like 32 ETH. If you're making 1.5 ETH a month this month, next month it'll be maybe 1 ETH (or less) and so on. In 2 years you'll be making basically nothing per month, definitely not 1.5 ETH. Believe me - I've done this with BTC mining and never made a profit.
In fact, you'll be lucky to break even. However, Genesis gets your money, makes an immediate fiat currency profit on the hardware and electricity costs that remain constant over two years. They can take that money and buy ETH with it, or whatever - even putting it in the bank or stock market works - and make quite the bundle. They even get to cream off a few percent on any ETH you earn further increasing their bank - all for free.
Caveat Emptor (Buyer beware)
Oh, and with Ethereum there is one more fly in the ointment to spoil your mining hopes - if it switches to a proof of stake system this kind of mining will no longer be possible. They mention that in their notes. At that point, they will try to hash anything else that is profitable but you could easily be SOL.
So I remain convinced that you should only be doing this (mining) if your costs are very low. If you have free or very cheap hardware and someone else is paying for the electricity then, by all means, have at it - you'd be silly not to. Just where that break even point is going to be for ETH I don't know - I have previously bought a $200 GPU and mined various coins, made back maybe about $200 at the time.
Most altcoins were only briefly profitable unless you gambled and waited years for LTC or others to come back in the recent bubble. But it's really not a no-brainer you'll make any money. Unless - and I'll say it for the third time - your costs are very low. If someone offered me 100MH/s for $100 hell yes I'd take it. Probably even at $500 it would be a no-brainer. But at $2800 - no thanks.
Home build costs
Following a guide like this: https://www.cryptocompare.com/mining/guides/how-to-build-an-ethereum-mining-rig/  will get you around 100MH/s with 5x Radion HD 7950 which go for around $200 each, plus mobo, case, RAM, PSU etc. You're probably talking about $1400 or more all in. Then unless you have free power you're looking at paying for 1kW of power, which depending on where you live, could be anything from $0.05/kWh to $.30/kWh (power usually gets more expensive the more you use per month). Let's say $0.15/kWh - that is about $110/month.
So let's say your rig costs $1400, your power costs $100/month and over six months that is $2000. You could either buy about 6 ETH now or hope that the 6--month yield is more than that. Supposing the yield declines by 1/3 per month. So that's about 1.5 + 1.0 + 0.666 + 0.444 + 0.3 + 0.2 = 4.4. At that time you're already yielding less than the cost of electricity per month unless ETH is over $500.
Now granted at the end of the six months you could turn around and sell your rig to someone else, or just keep mining. But eventually, the yield will be less than your $100 a month power cost unless ETH is going up by more than 50% per month (that could well be - do you feel lucky punk?).
Again, if your power is free (yay solar power!) and you have the money for hardware, or someone gives you hardware, and you have a room for a noisy mining rig (or rigs) then go for it. But if not, and you really believe in ETH you could just take all that money, plus $100 per month and buy ETH today.