You don't need particularly high density for storing your roof solar. Price is a lot more important, many people still use lead-acid for this purpose.

I re-read the article to be sure and here is that answer.

$15000 per 10 year old Tesla battery is the answer to the question, 'how much would a 10 year old Tesla battery be worth to a utility if used to supply peeking power?'

Ans: $15k

They get that number multiplying the going rate for peek power by the remaining battery capacity over 16 years.

Year 1: 0.0385 MW/days X $100 X 365 days = $1405

Over a 16 year investment period adds up to $15k.

But the batteries don't have a capacity of 0.0385 MW-days. They have a capacity of 0.0035 MW-days when brand new, or 0.0032 MW-days if degraded to 90% capacity. (It looks like they divided kilowatt-hours by 2 instead of 24 to get kilowatt-days.) Not only that, but it looks like off-peak power is about 70% as expensive as peak power, so it'll cost $70 per MW-day to charge the battery.

Year 1 (corrected): 0.0032 MW-days x ($100 - $70) x 365 days = $35.04

Edit: This assumes they're cycling the battery once per day. Presumably there's only one period each day when they can sell at peak and one period each day when they can buy at off-peak.

You're making a wrong assumption about what is being bought. (you aren't alone, everyone else in the thread is as well)

What's being purchased is 'standby reserve power per day 'not 'energy per day'

Digging it appears the promise being sold is. Will reserve of 2 hours of peeking power for $100 per megawatt _per day_.

That's where the divide by 2 comes from. because what you are selling is the ability to provide 2 hours of peeking power _if needed_. Not 24 hours worth.