OK - quite a bit of speculation on here. Let me ask a few pertinent questions as I work in this field and have assessed many Data Centres across the World.
1. For this method to work, we need to understand Google's business model. It can lose any single DC, (one in Belgium has no air conditioning at all - it may be this one in the article), and it's search facility is affected only minimally as every query goes out to multiple DCs anyway, with the fastest response appearing on your screen. If the batteries lose juice in an extended outage, then they probably don't care, which brings me to the next point:
2. Why do this, when it has UPS that can cleanly shut down its equipment in any outage greater than, usually, 7 minutes? No sane enterprise DC will risk this architectural model
3. Batteries are incredibly expensive and constitute one of the largest replacement items in an operational DC. Usually VRLA (lead acid) batteries will last around 8-10 years MAXIMUM, are around 90% renewable, but cost £'000,000s to replace when they all need swapping out - oh, and that's as long as you keep them within optimum operating parameters of 20-22 degrees C all the time. They also take up a lot of commercial space in a DC, will require good monitoring (ideally cell level) and can give off toxic fumes
4. If Google are considering Li-Ion, then the economics worsen and the sustainability equation moves out to the right (much less sustainable) and don't mention 'thermal runaway' risk
5. Generators have a lifetime of around 25 years minimum - some may even last, with excellent maintenance, up to 30-35 years and come up to a stable input voltage very quickly, which is why they are still preferred over fuel cells
6. The energy is not as 'clean' as Google would have you believe - when it purchases all the wind power for its DCs, this means everyone else is using non-renewables and that means getting their energy from the grid - whatever the carbon make up is of that. Therefore charging up its batteries is not 'free' nor 'clean'. If Google wants to make a difference to carbon-intensive energy use, it should invest its billions in 'greening' the grid for everyone (as an aside, you should look up the 'sustainability' of wind turbines - you may be shocked to learn they end up in landfill after only a short lifetime of use)
My honest opinion, having read the minimal information in the article, is that this is a 'concept' test that may suit Google but hardly anyone else - very similar to the Microsoft underwater DC it developed, which was also headline-grabbing but a really bad idea for many practicable, logistical and climate change reasons