It's just the cost divided by the storage power assuming a daily discharge of 80% of capacity for the 10 year warranty period.
That analysis is accurate too, but it's an extreme scenario (Edit - See below). Looking at the warranty document they referenced, In Australia, Tesla warranties the product to maintain 60+% of capacity after 10 year or 18000MWh, which is $3000/18000000kWh = $.16+/kWh. It's unlikely that every single powerwall will drop dead at the end of it's warranty period, so overall storage costs will likely be lower than that. Granted, the worst case scenario would be if someone purchased a powerwall and threw it off a cliff. That would be very expensive storage. ;)
https://naturalsolar.com.au/wp-content/uploads/2015/12/Tesla_Powerwall_Warranty_Australia_R1-04_en1.pdfDegradation also isn't a huge issue for NCA cells, but since heat has a substantial influence on lifespan, I can see why Tesla would have a relatively restrictive warranty in Australia. With that said, I'm not sure why you're resorting to ad-hominem attacks when a few minutes searching for NCA cell performance would have provided you with that information.
http://www.nrel.gov/transportation/energystorage/pdfs/45048.pdfData from over a decade ago (page 18) shows NCA cells can handle DoDs of ~70% and only lose ~2+% capacity/year as long their operating temps are regulated. Tesla, Toyota, and whoever else isn't using some special chemistry. They're using a battery chemistry that's been around for at least a decade, and has probably been improved over that time. If they were using something that died after a few years, they'd be bleeding money by offering to warranty these batteries at ~16+c/kWh. Odds are, the actual energy storage costs will be lower than the warrantied storage costs, and the cost of a cell replacement on an existing powerwall will be much lower than the cost of a new powerwall, which means storage costs going forward will drop.
Anyone can pay 10c, or 6c, or whatever at night from the grid, but with every utility co I've heard of, they also have to purchase electricity from them during the day, which can be far more expensive, instead of using less expensive power they generate themselves. Like I said before, this is primarily advantageous when someone can control their own production and use. If they absolutely positively have to power their "grow lights" at night, then a powerwall wouldn't work out for them. On the other hand, if they can bias their energy use towards day time power consumption, a powerwall + solar PV may be less expensive because customers that use less and can control their consumption still have to pay about the same amount for transmission, distribution, and admin overhead from the utility, but can actively reduce their self-generation/storage costs by going with PV panels and minimizing the amount of energy they use at night and need to store.
Edit - Now that I'm looking at it some more, that link you posted has some issues. The author's graph assumes the capacity warrantied at the end of each period will be the same during that whole period, which isn't accurate. If someone's powerwall has dropped to 85% of capacity (6.4kWh) at any time prior to the 2-year/4MWh mark, Tesla will replace or repair it. In fact, the author's assumption includes a powerwall that is only operating per the warranty spec 3 days out of the 3650 day warranty, which is, um... interesting.
So yeah, a defective powerwall that Tesla is never asked to fix, even though the defect is present through virtually the product's entire life, can be more expensive than a functional powerwall that is within the specs of Tesla's warranty. :D