Image Courtesy: Gemini

Key Points:

• Microsoft has laid off over 15,000 employees in 2025, signaling a significant workforce shift amid accelerating AI adoption.
• The company also released an in-depth study analyzing how AI impacts different occupations, based on real-world Copilot usage across 200,000 interactions.
• We compared Microsoft’s actual layoffs with the findings of its own AI research – mapping job cuts against AI applicability by function and task.
• Conclusion: Microsoft is clearly cutting where its Copilot flies – aligning its workforce strategy with the roles and tasks most affected by AI and offering a playbook for how AI disruption may unfold across industries.

The age of AI is no longer theoretical. It’s real, operational, and showing up not just in product roadmaps, but in company org charts.

Consider these headlines from just the first half of 2025:

• Over 10,000 jobs were cut due to generative AI adoption, according to Challenger, Gray & Christmas, making AI one of the top five drivers of layoffs this year.
• 40% of employers say they expect to reduce headcount as AI takes over tasks, per a World Economic Forum report.
• CEOs from Workday, Autodesk, CrowdStrike, Tata Consultancy Services, and others have pointed to AI investment as a key reason for recent layoffs.

And then there’s Microsoft—perhaps the clearest case of transformation in action. In 2025, the company has cut 15,000+ jobs while investing more than $88 billion in AI infrastructure. While Microsoft downplays AI efficiency as the primary reason for layoffs, however, President Brad Smith admits that soaring AI-related capital expenses are putting pressure on operating costs.

In July, Microsoft released a study:
“Working with AI: Measuring the Occupational Implications of Generative AI” – a deep dive into 200,000 Copilot interactions, highlighting which jobs are most exposed to disruption.

So, we decided to take a closer look:

Do Microsoft’s layoffs match the very roles its own AI research says are most at risk?
In this blog, we compare Microsoft’s own AI disruption insights with its layoffs in Washington state. 

In short: Is Microsoft cutting where its Copilot flies?

 

Microsoft’s Research: What It Reveals About AI’s Impact on Jobs

 

What Was the Goal?

Microsoft set out to understand how AI is being used at work and which types of jobs are most likely to be impacted.

What Did They Study?

They analyzed 200,000 anonymous conversations between users and Microsoft Bing Copilot (their AI assistant). The goal was to identify:

• What kinds of tasks people are asking AI to help with.
• How successful AI is at completing those tasks.
• Which jobs include those kinds of tasks.

What Did They Find?

• People most often asked AI to gather information and write such as emails, summaries, and reports.
• AI performed best at providing information, writing, teaching, and advising.

How Did Microsoft Measure AI’s Real-World Effectiveness?

To move beyond what people ask AI to do, Microsoft evaluated how well AI actually helps in real work scenarios. To do that Microsoft calculated an AI Applicability Score for each job – essentially, a measure of how useful or disruptive AI could be in that role.

The score is based on:

1. Coverage: How often AI is used in that field.
2. Completion: How successfully AI finishes tasks.
3. Scope: How much of the job AI can do.

The final score combines all three factors to estimate how much a job can be supported by AI.

What Does a High or Low Score Mean?

• High Score = AI is frequently used, performs well, and supports key parts of the job.
• Low Score = AI is rarely used or struggles to meaningfully assist.

Which Job Categories Are Most (and Least) Affected?

To understand broader trends, Microsoft grouped jobs into 22 broad occupational categories (SOC Major Groups):

Top 3 Job Groups Most Affected by AI

 

Job Group	AI Score	What AI Helps With
Sales and Related	0.32	Emails, research, client suggestions
Computer & Mathematical	0.30	Coding, data analysis, technical writing
Office/Admin Support	0.29	Emails, documents, data entry, scheduling

Other Areas Where AI Helps a Lot

• Community and Social Service
• Media and Arts
• Business and Finance
• Teaching and Education

These jobs rely on communication, writing, and information, which AI currently supports well.

Jobs Where AI Isn’t Very Helpful (Yet)

Job Group	AI Score	Why Not?
Farming, Fishing, Forestry	0.06	Hands-on field work, not digital
Healthcare Support	0.05	Personal, physical tasks (e.g., caregiving)
Construction & Extraction	0.08	Physical, tool-based labor

 

These jobs require physical presence, hands-on skills, or deep human interaction – not easily done by AI – at least not until physical AI, like humanoid robots, become viable.

Microsoft’s Job Cuts vs. Its Own AI Impact Study: Are They Aligned?

So far in 2025, Microsoft has laid off 3,160 employees in Washington state. For 2,290 of those jobs, we have a clear breakdown by role.

 

Microsoft’s layoffs in 2025 show a sharp focus on roles in engineering, program/product management, marketing, design, and data science – many of which its own internal AI applicability research flags as most exposed to automation.

Let’s compare the layoff data with the study’s findings in Table and the broader research summary.

Microsoft’s AI Applicability Study – Key Insights Recap

• Top AI-Impacted Occupational Groups (High Applicability Scores):

• • • Sales and Related
    • Computer and Mathematical
    • Office and Administrative Support
    • Community and Social Service
    • Arts, Design, Entertainment, Sports, Media
    • Business and Financial Operations
  • Roles involving writing, researching, evaluating, and communication tasks showed high satisfaction and task completion rates when supported by AI.
• Data analysis and visual design tasks showed lower satisfaction and scope, but AI still completes many of them effectively.

Microsoft 2025 Layoff Roles vs. AI Study Mapping

 

Microsoft Layoff Role	Study SOC Group Equivalent	AI Applicability Score	Comments
Engineering (993)	Computer and Mathematical	0.50 – 0.64	High AI applicability; AI now writes ~30% of code.
Program Management (412)	Business and Financial Operations / Admin Support	0.29 – 0.47	AI handles tracking, reporting, coordination.
Product Management (410)	Business and Financial Operations	0.47	AI helps with planning, prioritization, analytics.
Marketing (78)	Arts, Design, Entertainment, Media	0.25	Generative AI is strong in content generation.
Design (45)	Arts, Design, Entertainment, Media	0.25	AI assists but with lower satisfaction for visual design.
Data Science (26)	Computer and Mathematical	0.50 – 0.64	High AI capability in modeling, analysis.
Applied Sciences (25)	Life, Physical, Social Science	0.20	Mixed results; AI helps with research synthesis.
Business Management (24)	Management	0.14	Lower applicability; strategic decisions less automatable.
Legal Counsel (22)	Legal	0.13 – 0.42	Some impact via contract review and compliance.
Support (21)	Office/Admin Support / Customer Support	0.29	AI chatbots increasingly take over first-line queries.

So, is Microsoft cutting where its Copilot flies?

 

Clear Alignment with AI Findings

Roles like Engineering, Program Management, Product Management, Marketing, and Data Science make up the lion’s share of the cuts – and not coincidentally, they also score high on AI applicability in Microsoft’s research.

These jobs often involve structured, repeatable tasks like writing code, managing workflows, generating content, and analyzing data – precisely the kind of work that Microsoft Copilot and similar tools now handle with speed and accuracy.

In other words, Microsoft isn’t just talking about AI’s impact – it’s cutting where its Copilot flies.

But What About the “Low Applicability” Roles?

Here’s where things get interesting.

Some job functions that received lower AI applicability scores in Microsoft’s study – like Applied Sciences, Legal, and Business Management – were still hit by layoffs. At first, this might look like a mismatch. But dig a little deeper, and the alignment becomes more nuanced.

Applied Sciences

While the overall score for science-related jobs was lower, this is an averaged result across a broad category. Within applied sciences, AI is already effective at research synthesis, literature scanning, data modeling, and even hypothesis generation – especially in roles like research assistants or computational scientists. In those subfields, the impact is very real.

Legal

The legal profession received only modest AI applicability scores overall – but not all legal work is created equal. AI may not be arguing in court, but it’s already reviewing contracts, conducting legal research, drafting clauses, and summarizing case law. These are precisely the types of tasks being automated today. Legal support roles like paralegals and contract analysts are especially exposed, as they increasingly find their work augmented – or outright replaced – by AI.

That said, many legal professionals rely on specialized AI platforms like LexisNexis, Thomson Reuters, and Harvey AI, which are more tailored to legal workflows than Microsoft Copilot. As a result, Copilot’s role in legal automation may be more limited – but the broader trend of AI transforming legal support work remains clear.

Business Management

Strategic leadership may remain largely human-led, but day-to-day operational tasks – like budgeting, reporting, and internal communications – are already being AI-assisted. Some middle-management and support functions may be exposed to efficiency pressures, even if the top-line job title suggests otherwise.

 

The Bottom Line: Microsoft Isn’t Just Talking the Talk

When you line up the layoffs with the AI research, the pattern is clear: Microsoft is putting its pink slips where its Copilot is.

Most of the roles cut either rank high in AI applicability or contain subsets of tasks that are already being automated – whether it’s writing code, generating content, analyzing reports, or reviewing legal documents.

Yes, some of the layoffs are likely due to cost-cutting and restructuring, not just automation. But overall, the alignment between where Microsoft says AI works best and where it’s cutting jobs suggests the company is not just forecasting the future of work – it’s already reshaping it.

 

About JLA

 

Sangit Rawlley is a Senior Partner and AI Practice Lead at JLA Advisors (www.jlaadvisors.io), a boutique consultancy helping organizations navigate the complex and fast-moving world of AI. From workforce strategy to automation roadmaps, JLA’s AI practice specializes in turning AI disruption into competitive advantage.

If your business is wrestling with the same questions facing companies everywhere – Which roles are AI-ready? Where should we invest, augment, or restructure? – JLA can help.

 

📅 Let’s talk: Schedule a call to explore how AI can drive meaningful, measurable impact in your organization.