any chance to calculate the estimated ROI ?
It is not easy to give an accurate picture as there are many variables that affect this. At current difficulty, ROI would be achieved in a month. But three months into the future, difficulty will be higher. How high exactly depends on a number of assumptions.
e.g. organofcorti.blogspot.com forecasts that it will increase about 7 fold by end of July. In that case, ROI will take about six months. With overclocking, it will be four and a half months. He of course assumes that Avalalon batches 1,2 & 3 will have shipped by then, BFL will have shipped and ASICMiner will also have about 100Thps by then. If any of these assumptions do not materialize then difficulty will not increase 7x and ROI will be achieved much earlier.
You cannot estimate ROI with a static difficulty. The hashrate will continue to increase every two weeks by a large factor - and likely at a ridiculous rate once the raw Avalon chips are available. CONSERVATIVELY estimating you starting your mining the second week in July, if you somehow ordered today, got your chips in the 10 week shipping schedule projected by Avalon, and did all of your assembly, you are looking at a difficulty of no less than 35,000,000 and likely closer to 50,000,000.
If you assume the lower 35 mil difficulty, and your overclocking is stable to the max of the chip (3,000,000 MH/S for your 10,000 chips), you will make about 300 bitcoins each week for the first two weeks for a total of 600 bitcoins.
After those 2 weeks (which will be less than two weeks because the network will be solving blocks faster than every 10 minutes as more Avalon chips come online) the difficulty will jump by another 30% at least... so your next two weeks at 45,500,000 you'll make about 230 coins per week (you will now be at 1060 coins).
The following two weeks will be close to 60,000,000 difficulty... for 175 coins each week and a total of 1410 coins after six weeks of operation.
Extending this estimate formula out to 3 months gets you to just over 2040 bitcoins in 3 months.
No one can know what the difficulty will be, but these projections aren't out of line and many think the hash rate will double each month until the end of the year and beyond. I just think it's not reasonable to expect a ROI of 3500 coins within the next 10 to 12 months (or possibly ever) on 3000 GH/S even under the "favorable" conditions outlined above.
If you aren't starting until mid-July at the absolute earliest and you are trying to ROI in 3 months, you really can't have your all-in pricepoint set above 1400 btc. Even at 2000 btc your product would be more palatable - but 3500 is quite high and I wouldn't expect to see your coins back for quite some time.
This calculation completely omits the extra 15% admin fee, which will only hurt the numbers above when included.