Read together, the week's stories tell one story from different angles, and it's about dependency.
A frontier model went offline by external decision, and the company admitted it was serving essentially no traffic during the period. A large institutional buyer began testing competitors to replace that model in sensitive systems. Corporate products launched the same week shipped supporting three vendors from day one.
The three stories describe three phases of the same lesson: the supplier fails, the customer looks for an alternative, the market starts requiring alternatives by default.
The detail that usually goes unnoticed is the middle of that sequence. A customer who migrates out of necessity adapts their code, and once adapted rarely switches back simply because the original supplier returned.
For operations of any size, the practical lesson doesn't change with scale: keep a qualified alternative, tested under real load, before you need it, and accept that it costs a little more day to day. That's the price of not stopping.