⚠ The Great UntanglingLow threat

The threat to integration depth is that the tangle which binds customers is also a burden they would dearly love to be rid of, and a whole industry has grown up promising to untangle it. Integration platforms — the likes of MuleSoft, Zapier, and a dozen others¹ — sell the ability to connect any system to any other through standardized connectors, which in principle makes the underlying stack more swappable: if the connections run through a neutral layer, the thing underneath becomes a component you can replace.

Artificial intelligence raises the stakes, because the hardest part of any migration was always understanding and rebuilding the undocumented tangle, and AI is increasingly able to read a legacy system, map its dependencies, and generate the code to reconnect it elsewhere. A moat that depended on no one understanding the whole weakens if a machine can understand the whole.

In practice, though, these tools more often deepen Microsoft's position than loosen it, because the neutral integration layer usually ends up integrating everything to Microsoft, whose products sit at the center of the enterprise anyway. And live, business-critical integrations are not migrated merely because a tool makes it theoretically possible — the risk of breaking the Monday report is too high whatever the tool promises.

Call it low-to-moderate. The untangling tools and AI mapping are real and improving, and over many years they could make deep stacks more portable than they have ever been — but the enterprise's aversion to risking what already works is itself a moat, and a customer rarely rebuilds a thousand working connections just because it has become possible to.