From the producer side: web applications are faster to prototype, develop, publish and monetize than desktop applications. This is also a massive advantage for the average application, cost wins.
The main advantages a native desktop application still has are performance, native look & feel and native feature support.
Purists will argue these native features are super important (they are in particular cases), meanwhile economics will keep pushing towards more web and electron apps. Such is life.
Also the users local environment becomes less of a factor.
Getting a user to install a piece of software, is a large point of friction. To get them to do that, you have to override all the negative experiences users have inevitably had with installed software (breaking their system, malware, abuses like spying on your machine, compatibility, and so on).
With a Web application, you can often immediately hand a user a demo key to dive right in and begin trying the product. Depending on the product, in some cases you can present a demo live on the site, no key or signup necessary, the user can instantly begin fiddling around.
You can't do that with a desktop application, you'll have to convince the user to download something and install it, even if it's just a demo.
A Web app lowers the bar to first use, that moment of convincing someone to actually try out your product. This speeds up and simplifies your distribution. Not having the product installed in thousands of locations, on thousands of different machines with different specs, dramatically lowers your support burden as well: there's one live version and it's entirely under your control. You can train your support team for that homogeneous case.
If you have a desktop app that in any way interacts with your backend, you need to take into account active client versions before upgrading your backend. With web this problem basically disappears.