JohannPopper Web browsers are far from standard or reliable as a platform. They integrate poorly with the OS they run on and run some of the worst client code I have ever seen. You can't even guarantee compatibility between versions of a browser, let alone different browsers entirely. And I should know, I've done web development for a living.
And this whole notion that everything is moving server-side is stupid. Well designed web services push as much as they can onto the client without hurting the user experience. Why? Service providers don't pay for the compute performed by the client, they pay for what the servers do. You want the servers doing as little as possible to provide the service: that's how you make money. This is especially when the service you provide is ad-supported. In fact, the whole of WebRTC is designed to take the server out of the equation. The servers are basically only there to hold on to text messages and event logs for the future, which is really cheap to do. Video and audio? Expensive as hell. WebRTC is designed to connect clients as directly together as possible. It's the ultimate example of pushing compute onto the client.
As for the whole client-server dynamic, it's a trend in the industry that cycles back and forth. We started with academic and business mainframes then pushed compute out to microcomputers so that terminals wouldn't eat up valuable mainframe time. Those computers became household appliances and then we pushed back to the data centers for the original Internet. Personal computers became portable and smart phones were invented, pushing compute back out to the clients again. Then virtualization was invented and everything got pushed out to the public cloud because not everyone can afford to own a datacenter or pay for COLO hosting. Now the amount of data produced is so huge that we had to invent the field of edge computing to preprocess everything and compress the information we care about before it finds its way back to the datacenter. A datacenter now needing insane amount of computer power to handle all of this data for anything meaningful. Leading to ML software which trains neural networks which we then push out to client devices so that they can execute them instead of the server. And this trend will go on and on and on....
You may have drunk some of that Kool-Aid, but I live in the real world where companies invent new technologies so that the services they already provide continue to make them money. In that world Client-Server is a pendulum and you just hope you guessed correctly which direction its currently heading.