I have measured some of those "300 cycle" cells and they perform way better.
I guess they just leave some leeway for process variance, i.e., don't want to guarantee more than 300 cycles.
On the other hand, exotic cells that are guaranteed to very large cycle ratings (LFP and LTO chemistries; thousands or tens of thousands of cycles) have the tendency of underperforming when the manufacturer is too eager to market their fancy product and do not understand all the aging mechanisms involved.
Unsurprisingly, the optimal case is to use cheap enough mass technology, but commit to buy in hunderds of millions $, then work in close cooperation with the manufacturer to find how that 300 cycles can be made 1500 cycles without significant or expensive changes in the product. This is what all the EV manufacturers are doing. No magical special EV chemistries, but not exactly the consumer cell product lines either.
Tesla started with bog-standard off-the-shelf consumer cells in Roadster, IIRC. That was revolutionary; others were working with battery manufacturers too early with too little budget. Time was not ripe for that kind of cooperation or "custom EV chemistries" then. Using COTS consumer cells first was a bold move but the right one. Then Tesla went on and started to refine the process, so the cells stopped to being exact same consumer cells, but were still very close. For example, Panasonic NCR18650B for the early Model S, IIRC, where the major modification was the removal of PTC cap but there could have been other minor modifications as well; if nothing else, more careful screening.