These are real risks to these companies.
Your in-house teams can build replacements, it's just a matter of headcount. With Claude, you can build it and staff it and have time left over. Then your investment pays dividends instead of being a subscription straight jacket you have to keep renting.
I think there's an even faster middle ground: open source AI-assisted replacements for SaaS are probably coming. Some of these companies might offer managed versions, which will speed up adoption.
Lets take Figma as an example, Imagine you have 1000 employees, 300 of them need Figma, so you are paying 120k per year in Figma licenses. You can afford 1 employee working on your own internal Figma. you are paying the same but getting 100x worst experience, unless your 1 employee with CC can somehow find and copy important parts of Figma on his own, deploy and keep it running through the year without issues, which sounds ludicrous.
If you have less than 1000 employees it wouldnt even make sense to have 1 employee doing Figma
I mean in an example that almost happened... "you are paying 120k per year in Figma licenses, Adobe buys it, you are paying 500k per year in Figma licenses"
At least up until the point of vibe coding it was still worth the SaaS provider charging at least as much if not slightly more than you doing it yourself because most businesses weren't going to anyway.