As long as Tesla is battery cell constrained (e.g. next 5 years or longer) every kWh of cells they sell needs to be attached to a certain amount of profit. This is why Semi is not going to ship in volume for a while.
There’s rumor Tesla will build a cheaper compact car for China and EU. If you allocate 45kWh of cells to a car which sells for $20k and earns $4k of profit, versus 75kWh of cells to a car that sells for $50k and $10k of profit, all else being equal (20% GM) you want to sell more of Option B because the battery cell is your manufacturing limit and profit per kWh is $133 vs $89.
Or to put it another way, a car that clears $4k for Tesla can only have a 30kWh battery, or else selling that car drives down profit.
Tesla is rumored to be planning on making their own chemistry cells at GF4. I expect we will see real volume in their next-gen chemistry by the end of 2022 or mid-2023, at which point their annual production run rate will likely be greater than 1.5m vehicles requiring 100GWh of batteries per year. Currently they are closer to 35GWh, and IMO scaling the battery production is by far the hardest part.
They have started using LG Chem cells in China, whereas up until now they were only sourcing cells from Panasonic. This is because it’s simply not feasible to produce the needed cells for GF3’s 150k/year run rate at GF1 and ship them overseas. Although that’s what they did for the initial China production run.