The Web Speech API is still in "Draft" spec, but Mozilla have been working steadily on it and finished the work for SGML parsing back in January. There are actually flags you can change to enable it on most platforms, its just they aren't considered "stable" yet.
Mozilla's DeepSpeech engine is also not that far off from reaching a level of stability. You can help them with it here [0] if you're interested.
The reason Mozilla has not text-to-speech is because the guy behind RNNoise and LPCSpeech (JM Valin) has been bought by Amazon so that they can build even more privacy invasive Alexa & friends device.
So yeah, in a way this an issue of money, but not because you can't afford GPUs & Data, but because you can't buy the right people
That is part of the problem.
With smaller projects you are at risk of the "hit by a bus" problem, where major/complex technology is dependent on a single individual or a very small team.
Either way, people will choose the free option rather than the option that requires them to pay even a nominal fee, particularly for software that people have gotten used to getting for free.
I've been waiting for a WebspeechAPI implementation in Firefox for many years now but the only reason why I don't think it's happening is because maybe Mozilla lacks the funds to train the models for speech-to-text/text-to-speech, etc.
This gap will grow more and more as new machine learning features are introduced (e.g. natural lang page searching, better translation, etc) which would be very easy for Google to add to it's browser but very expensive for open source projects.