What Changes Should We Expect When Moving AI and Development Workloads to Browser-Based Apps?

We're considering a transition in how we deliver AI and development environments. Rather than providing users with complete desktops or virtual machines (VDIs), the idea is to enable access to essential tools like Jupyter, IDEs, terminals, and dashboards directly through a web browser. This approach would eliminate user-managed operating systems and persistent VM images, as everything would operate on pooled compute resources with centralized storage and ephemeral application sessions.

Key points we're hypothesizing about include:

* A significant drop in issues related to environment drift, driver mismatches, and post-update failures.
* Simplified compliance and patching processes since infrastructure teams would oversee the entire runtime stack.
* Better resource utilization through pooled GPU and CPU capacity versus allocations tied to individual workstations.

I'm particularly curious about the following considerations:

* Would tickets related to environment and setup actually reduce, or just shift to new categories such as browser or identity-related issues?
* How would single sign-on (SSO) and role-based access control (RBAC) operate effectively across browser-only environments, especially when using shared resources with different workloads?
* What challenges could emerge regarding data residency and data loss prevention as all operations shift into the data center or cloud?
* How do developers feel about a completely browser-centric setup in terms of latency, performance, and overall control?
* What early operational metrics would best illustrate the return on investment to leadership, like reductions in support tickets or improvements in infrastructure utilization?

This isn't a sales pitch; I just want to ensure we're thinking critically about these assumptions before diving deeper into this transition.