These are fair points and I am not defending TimeGPT in any way in fact it was interesting to see your previous article showing it underperforms compared to TIDE (I have posted TIDE many months back when it was published saying it was a good model).

My comment is to highlight that it is important to make readers aware that Amazon Chronos under-formed simple stats ensemble. It is irrelevant whether Nixtla has a stake in the game or not - they have been professionally benchmarking various models including from Amazon for at least 2 years before TimeGPT even appeared including showing previous models from Amazon like DeepAR are not competitive with stats ensebmbles.

Why is it important to highlight this? Consider Facebook prophet - a dysfunctional model that spread through data science ecosystem because no one was critical of it.

That is just my perspective, you guys are doing great job!