But do you know precisely how insecure it is? Anyone who can MITM the connection can break it in under a day with decade-old cracking technology.
The MITM step isn’t a huge hurdle. There are several available methods, all of which are reduced to freely-available tools. Since the best method for avoiding MITM is to encapsulate the traffic in a properly-designed secure tunnel — TLS, SSH, proper VPN, etc. — the argument that MITM isn’t a concern falls in on itself when it comes to questions such as this thread’s.
PPTP is really really really bad!
a large number of my clients’ embedded devices require VPN
How about you draw out the network design, and let us come up with a workable migration plan for you? The only way PPTP is your only feasible option is if you’re about to go out of business. By putting your trust in PPTP, you’re risking that already.
my clients are using dedicated embedded devices. No one will develop a SSTP stack for it for free
Ship each site a pre-configured hEX or similar to terminate a proper VPN tunnel. For instance, a site-to-site WireGuard tunnel, which can be port-forwarded through a NAT layer. That’ll solve your PPTP problems for very little money compared to the company-ending liability risk you’re taking on by not doing something like this.
If the devices won’t talk over anything but PPTP, you can still tunnel PPTP through the outer tunnel. It’d be inefficient, but it’d work.
I have to suffer full window of red messages which is really annoying and distracting.
How much suffering, annoyance, and distraction do you suppose you’re in for if someone decides to start cracking your PPTP tunnels?
You don’t get a choice between zero annoyance and some annoyance. You only get a choice of which bag-of-annoyance you’re willing to pick up. Passively avoiding the choice is still a choice, because it merely means someone else gets to decide which bag-of-annoyance to hand you.
Time is not the problem. Money is.
Do you suppose cleaning up after a successful attack will be cost-free?