They're currently rated for 20 years of usable life, so you'd replace 5% of them a year. They're supposedly 95%+ recyclable, it's just not currently economically viable to do so since only cobalt and nickel are cheaper to reuse than to newly mine.
A project I had a colleague working on where the batteries are critical to preventing blackouts since they will result in mothballing of other infrastructure indicated the utility was budgeting for a 5 year lifespan on the batteries.
It's what Tesla puts on their front page marketing[1].
But if you want more detailed breakout, there's this[2] report that you can dig through which states:
> A range of cycle estimates was provided throughout the literature for lithium-ion of up to nearly 6,000 cycles with lower DOD (DiOrio et al., 2015; Greenspon, 2017). The analysis conducted here estimates that lithium-ion LFP can typically provide 2,000 cycles at 80% DOD, while NMC systems provide 1,200 cycles for the same DOD, due to positive electrode dissolution and associated increased capacity loss at the negative electrode. In the next phase, more detailed cycle life data for LFP and NMC chemistries will be obtained. For example, based on 70% capacity at end of life, lithium-ion batteries have demonstrated a cycle life of approximately 8,000 cycles at 80% DOD (R. B. Wright & Motloch, 2001). The calendar life of lithium-ion batteries ranges with some stating > 5 years or as high as 20 years (R. B. Wright & Motloch, 2001) and others in the range of 5-15 years (Dubarry, Qin, & Brooker, 2018). This report estimates a 10-year calendar life at 80% DOD, also assuming 5% of that time will also be allocated to downtime. A cycle life of 2,000 cycles for LFP and 1,200 for NMC is assumed with a 5% increase in total cycles each by 2030.
So with the right chemistry, assuming one cycle a day, assuming 70% depth of discharge is acceptable, 8000 cycles is 21.9 years.
