The company never said their downtime was related to the leap year afaik, that was some random twitter thread looking at requests on their browser that had datestamps from the future (which could have other explanations). Unless they're rolling their own date library (which is very unlikely), the code should be using the system time of whatever computer the software is running on, meaning it would be extremely unlikely that the leap year caused a bug.
You are correct in that it's more likely their system just can't scale.
You are correct in that it's more likely their system just can't scale.