The "problem" is those factories are going to spit out immense cash flows shortly after construction. $24b is just the 2024 estimate. With ever increasing efficiency, FSD and just operating leverage from scaling up, cash flows could very well be $100+ billion by 2028.
I mean, I recall Ron Baron estimating that Giga Shanghai paid for itself within like 5 months of construction completion. That's an eye-popping IRR of roughly 200% after accounting for the fact that construction and low rate initial production took about 2 years.
To a first approximation, the only way Tesla would be able to reinvest all cash flow would be if
production capacity investments are growing at the same rate as their capital from previous investments. Until about 2020, Tesla's ROIC was less than their production investment growth rate, and this is the main reason why they accumulated debt and had detractors claiming it was a structurally unprofitable pyramid scheme. But now, with Tesla's incremental ROIC on new factories in the 100% to 200% range, Tesla's 50-100% growth in manufacturing capacity will be unable to keep up.
Thus, if Tesla is not going to pay a dividend nor build up a multi-hundred billion cash balance in this decade, then they must have something else planned. A rapid robotaxi fleet buildout, where Tesla rather than consumers pay the upfront costs, is the most reasonable capital sink I can imagine. If we're aiming for let's say:
- A fleet of 10^8 AEVs
- By 2035
- With unit cost of $50k on average (this includes Semis and Vans/Minibusses dragging up the average)
Then that will take about half a trillion dollars of investment.
Another huge potential capital sink--I just realized--is investing in their own solar + battery projects as a virtual power plant operator instead of selling the batteries to someone else. This could also suck up on the order of 10^12 dollars. Same kind of logic: Tesla basically buying their own products with their own money and then sell the services over time. Tesla knows their own product and will have super cheap capital, and so could make arbitrage profit off this information asymmetry.