In the run-up to I/O 2019, Google announced that Googlebot is now running the latest version of Chrome (74) and will continue to be kept up to date, and there was much rejoicing.
Transpiling is basically transforming modern JS functions in to basic JS (1 line of code might become 50) and polyfilling is the process of emulating functionality missing in a target browser. Stackoverflow handles the answer pretty well.
Now it should just all work for Googlebot.
Anyone that has been involved in SEO for a while will tell you that you need to take announcements like this with a pinch of salt. This isn’t having a go at Google, updating their rendering to Chrome 74 would have been a major undertaking worthy of praise and they themselves would recommend you still test, test and test again.
So, make sure you test your rendering using…. erm we have a problem! At the time of writing, none of the tools are updated to Chrome 74 but Zoe Clifford from the Google rendering team mentioned that is on its way shortly during the recent I/O talk (at timestamp 6:22) on the subject so keep your eyes peeled for news.
Also, it is worth noting that the current Googlebot user-agent hasn’t changed. This is deliberate to give people time to modify any systems that might use a hardcoded user-agent for things like dynamic rendering. This was also covered in the I/O talk at timestamp 5:52.
Here is the real kicker. Why were you transpiling / polyfilling in the first place? If you were doing it for Googlebot solely then in theory you should be able to stop (test, test and test again).
If however you are doing it for users with older browsers, nothing changes, you still need to transpile and polyfill for them.
Can I ditch my prerender / dynamic rendering?
Here, it is important to understand why you are using this in the first place. If it is solely because Googlebot couldn’t render your site correctly, then by all means change things after testing.
This update to Googlebot’s rendering capabilities does not change the fact that rendering is expensive and therefore carried out as a second (often significantly delayed) wave.
Zoe mentions this at timestamp 18:24.
Zoe and Martin cover this very well in a simple form at timestamp 7.25.