Microsoft 4Q26 preview: Stagnation or breakout?

研报英文原文证据摘录

Microsoft 4Q26 preview: Stagnation or breakout?

rowth was good a lot of the added capacity went to supporting 1st party app growth (O365 Copilot paid seats

grew 33% QoQ last quarter) and toward building models (Microsoft is now delivering competitive focused models)

and memory price increases were a bigger factor than we suspected (3Q26 earnings). While the Copilot beat was

important investors want to see Azure growth accelerate.

• Adding powered SQ FT in datacenters takes time. Therefore, one plans and builds datacenters years ahead of when

they will be able to equip the datacenter and drive revenue. The fact that almost a year ago Microsoft started to

slow their build-out (shifting from ~60% of CAPEX to 33% of CAPEX) is an important statement - it means that

management believes that they will have enough SQ FT (the current limiting factor on driving Cloud revenue) to meet

ongoing demand likely within a year to at most 2 years. Meanwhile, they are leasing more datacenters, renting capacity

from neoclouds and even renting CPU capacity from Amazon (Link). In addition, Microsoft’s peers are pivoting to

renting to other Cloud / AI providers creating an interesting opportunity for Microsoft.

• Another example of Microsoft’s pivot to add capacity occurred yesterday. Microsoft (Mistral / Microsoft expanded

partnership) was expanding their partnership with Mistral, the European AI lab. Under the agreement Microsoft

would be using Mistral datacenters for some Microsoft Azure workloads in Europe. This would both create another

Microsoft European sovereign Cloud offering and also increase Microsoft datacenter capacity in Europe without

increasing CAPEX. While Microsoft’s gross margin on this revenue would likely be lower there would nominal if