Microsoft is fast becoming a fully AI company.

Yesterday Microsoft announced its decision to voluntarily buy out 27% (60,000 people) of its workforce as it moves further towards an all-in investment in AI. While it isn’t clear to me if that means they’ll be replacing a significant portion of jobs with AI, that seems like a strong likelihood.

I find the timing of the announcement to be very interesting given that only two weeks ago, a developer using nothing but AI tools completely reproduced the entirety of the Microsoft Office 365 suite with less than a week’s worth of effort. Called WordCraft, the software the developer released for free on Github is a complete clean-room implementation of Microsoft’s software. I’ll do a full discussion of this in a follow-up post but a clean-room implementation differs greatly from simply reverse engineering a piece of software. Courts have long protected the ability of developers and engineers to produce such offerings and the legality of WordCraft seems secure for now. The number of downloads isn’t known but the Github repository has been starred 16,300 times and there are more than 4,500 forks on Github, alone. Considering that Microsoft owns Github, it wouldn’t surprise me if it hasn’t been ported to other project hosting services as well.

This followed on the heels of the same developer’s release of the entire Adobe software offerings such as Photoshop, Illustrator, and Premier—named ArtCraft—that quickly spread around the globe. Another major release by the developer, the proverbial “icing on the cake”, was a Github repository containing the entire set of AI tools he used to perform this work, along with basic instructions for anyone who would like to do the same with other pieces of software.

To say this has had a chilling effect among the software giants would be an understatement but it is still too early to say what the full effect will be.

In trying to understand Microsoft’s recent motives for downsizing its workforce, it seems clear that a move towards more AI was already in the works. They have invested hugely in it over the past few years and it seems to have now become their top revenue-generating segment. It is also the core underlying technology behind everything they do moving forward.

Microsoft’s financial ecosystem has three main reporting segments, listed here in order of top-revenue generation for FY2026 (ending at the end of June). [Microsoft’s letter to shareholders of its Annual Report 2026]

1. Productivity and Business Processes: $139.9 billion

Includes Microsoft 365 commercial offerings ($102 billion) and Microsoft 365 consumer versions ($9.18 billion)

2. Intelligent Cloud: $137.7 billion

Microsoft’s fastest-growing segment, already expected to surpass their Productivity and Business Processes for FY2027. It includes their search technology but more importantly, their entirely new backbone infrastructure. They have already moved towards an all-AI backend that can produce software moving forward. For example, their AI tools have also ported the Microsoft 365 suite to Rust and has further automated development and deployment, obviating the need for human developers in many areas.

3. Personal Computing: $54.2 billion

This is the traditional core revenue stream Microsoft started with so many years ago. It includes its Windows operating systems as well as the XBox platform and other personal computer hardware offerings. It has also been re-aligned to grow off of Microsoft’s new AI infrastructure.

So looking at the financials alone, Microsoft took in $331.8 billion in total revenue for FY2026 with 42% of that coming from their productivity and business tools. Of course, no business is going to move away from Office 365 anytime soon, even if it were a fully compatible, non-beta release version. The consumer response, however, could indicate some trouble even though it represents only 2.7% of their total revenue. This is the segment of the market I want to pay attention to.

So far, Microsoft’s biggest money-maker has now been made freely available thanks to AI. Things look worse for Adobe, since everything they make revenue from has been released for free. Finally, the CAD giant AutoCAD has been similarly hit by the freely available CADCraft release.

You can see where this is going.

This “stick it to the man” type of software deployment isn’t the only use to which these AI tools are being applied. To choose one example, look at the software that runs on your new (within the last decade) flatscreen TV.

Remember when LCD TVs appeared and cost a fortune? Today you can buy a behemoth 85-inch screen TV for under $400. These were the $10,000 units circa 2008. Why have they come down in price so fast? The reason is because of the software that is installed on these units and the data that this software collects and sends back to the manufacturer. Advertisers have paid dearly to have apps installed on the TVs and they are paying for the data that the TVs collect not only about what you watch, but also about the other devices you have in your home and anything else that can be found out about your household. The fact that these TVs spy on us even when they are turned off was the subject of a recent interesting video.

There’s now a growing movement of folks who are replacing the operating system on their TVs with ad-free, spyware-free customizations. Using the AI tools mentioned above, this has become a simple and relatively painless process. You get the functionality of a clean Android-based TV box, for instance, with none of the bloat you do not want, a UI tuned to your liking, and no data collected about you and sent to an evil entity.

Another use case which appeals to me is the ability to reimplement older software which is no longer being supported at all, which I cannot even purchase if I wanted to. This is the case for a laser engraver driver that for now only runs on an old computer that I have spent hours keeping on life-support. I would love to reimplement the software in order to run it on a newer, and cheaper Linux machine. There are many examples of hackers out there doing just that for old binaries.

AI has drastically changed the business model and practices underlying software development. That is worthy of a whole other discussion. I’ll end with a quote from a reviewer who made a video about ArtCraft, the cloned Adobe tools. In addition to being a developer of photography-related software, he’s also a student.

“I have been studying computer science for like seven years and I’m just watching the industry fall apart. The job that I’ve always wanted my whole life, it’s gone.”