I designed Microsoft's EA channel in 2001. It's being dismantled in 2026 (brendanoconnor.net)
The piece walks through what the original architecture was, why it worked (Microsoft's volume licensing commitments nearly tripled in the two years following launch, $1.9B to $5.5B), and why the 2026 unwind is structurally different. The 2001 transition gave partners a defined advisory role with a sustainable economic model. The 2026 transition does not.
Happy to take questions on the architecture, the design decisions, the pieces that worked and the pieces I'd do differently, or how this maps to adjacent industries. Insurance brokerage M&A is currently following a parallel consolidation pattern, which is interesting because insurance was one of the two industries I studied when designing the ESA model in 1998.
If any specific claim needs sourcing, flag it and I'll respond with primary-source citations.
Brendan
9 comments
[ 0.27 ms ] story [ 46.3 ms ] threadCan we get TL;Don't Have MBA summary?
I fail to see how this is a win for the vast majority of folks impacted by the licensing process...
- There was no option to activate a specific set of machines sitting in a particular OU or AD group, which means you'd have to manually import machines that you want to activate - and you can't do a bulk import from a simple text/CSV file either.
- VAMT does not automatically retry failed devices, or devices which were offline. Trying to find these devices and manually activate them via VAMT is also not feasible, especially if you have a large fleet.
- Pretty much no useful reporting capability especially if you want to track failed/offline devices.
I've had to end up hacking together a bunch of PowerShell scripts and scheduled tasks to make up for the lack of functionality - something I shouldn't have to do. And it wasn't nice that the VAMT module isn't compatible with PowerShell v5.1, which means I needed to run my scripts under PowerShell v2...
And please don't tell me that the answer is "cloud". Cloud doesn't work for everyone, and I'm sure MS have plenty of Fortune 500 customers or critical organisations (like government agencies) that are still on-prem only.
The difference in the writing is structural. It sounded too elegant. The same outcomes produce different paragraphs because the words are attached to different incentives. LLM voice is a leading indicator that the work is inconsequential.
Reading further gave me the verified picture I needed. The mechanics held up. Most content creators won't last long enough to be both supported by their sponsors and reblogged by adopters. Claude's response to OP will be direct: hold the line.