I can vouch that delivering quality, scaled, enterprise video conferencing is a much bigger technical challenge than it seems after, say, a weekend or two hacking with WebRTC in a hobby capacity. It combines the challenge of supporting a huge landscape of differing hardware devices with that of supporting a huge landscape of differing network environments. There's a near-infinite combination of corner cases that can cause problems and handling them poorly creates the impression of severe instability for the users who experience them. The nature of video conferencing/collaboration software means if 1 user's environment makes the experience unusable, the value for everyone who needs to collaborate with them is also seriously hampered.

That said, when we started dogfooding our own conferencing product it was partly because Slack consistently had issues dropping calls after 5-10m of group video. I'm kind of surprised to read such similar complaints 18 months+ on. Presumably it's harder to solve those problems at mass market global scales, we're pretty vertical specific.