As far as I know, the strongest argument for tabs is provided by Elastic Tabstops. And this is the strongest argument against that. However, it's not entirely clear whether the performance really can't be improved, because no one has put that much effort into optimization.
Anyway, I think the big picture of spaces vs. tabs is often muddled because its scope is arbitrarily limited. For some, it might be insightful to take them into their respective logical conclusion.
The "spaces camp" values the simplicity of plain text. They like having this grid of cells they can draw things on with glyphs, which are laid out equally for everyone. The contents of the grid are then parsed into AST by the programming language of their choice, or transformed isomorphically by the formatter of their choice. To this end, things like syntax highlighting, type hint, indentation marker, ligature, etc. are not the first-class features of the code.
The "tabs camp" values the extensibility of rich text. They like parsing code style features like indentations and brackets as markups, of which style is determined by project and editor configuration, like how HTML style is determined by CSS and browser. This separation of style and structure, done consistently, would leave nothing for the formatter to do.
Both options have their own upsides and downsides. Plain text is easier to develop tooling for, and it can be sent anywhere without causing compatibility problems. Rich text, on the other hand, can better adjust to different people with their different visual preferences.
For now, the "spaces camp" have successfully captured the "don't care camp" by the virtue of requiring less editor configuration. Meanwhile, the "tabs camp" have struggled for the same reason XML did. Even though the expressive power of rich text is useful in some specific contexts, the amount of upfront and continuous investment required has made it unattractive to most people. Ultimately, the situation was not about which paradigm would result in a higher productivity, but rather "perfect is the enemy of good".