Most layoffs that get called "AI layoffs" are really concentration stories. Too much of a company's revenue, or too much of a person's career, rests on one thing. You can reduce that risk for yourself by spreading three things: your skills, your sources of opportunity, and your visible proof of work. Here is how I think about it, using this week's Innodata news as the example.
What happened at Innodata
On Monday, 26 October 2026, nearly 550 employees of Innodata India were reportedly laid off, according to India Today. Cuts were also reported in the Philippines. Employees described laptops being logged out and IDs deactivated during the workday, followed by termination emails to their personal addresses. NDTV carried similar accounts.
Innodata is a US-based company that helps technology firms build AI systems through data labelling, model training, evaluation and safety testing. As of the time of writing, the company has not publicly confirmed how many people were affected.
A termination email reviewed by India Today said the job was ending "due to the ramp-down of the business assignment for which you were engaged" and that the decision was "not related to your individual performance."
A follow-up India Today report said a WhatsApp group of affected workers had grown to 617 members, and that Innodata's shares fell about 25% on Nasdaq the next day.
Two ways to read the same news
The story went viral on X, and much of the reaction framed it as AI eating jobs. It is an understandable reaction. These were people helping to train AI systems.
But a second reading appeared quickly. CA Vivek Khatri wrote: "This is not an AI layoff story. It is a one-client story."
The numbers support that reading. Innodata's own 2025 annual report (Form 10-K) says one customer accounted for approximately 58% of total revenue in 2025, up from about 48% in 2024. The filing lists the loss of significant customers as a risk.
To be careful: Innodata has not said which client, if any, was behind the ramp-down. Some employee posts on Reddit, quoted by India Today, mentioned lower volume from a major client, but that is not confirmed. What we can say is that the company's own documents describe heavy dependence on one customer, and the termination email describes a project ending.
Why this matters even if you never worked at Innodata
AI is changing work. KPMG's 2026 India CEO Outlook, reported by Business Today, found that 52% of the 125 Indian CEOs surveyed plan workforce reductions in some areas, while 77% are hiring people with AI and technology skills. Both are happening at once.
But "AI took the job" is often too simple. In many cases the more useful question is: how many things does my income depend on? A company with one giant client is fragile. So is a career with one employer, one skill and no public proof.
That is career concentration risk. And unlike AI, it is something you can actually work on this month.
The three-part framework: spread your risk
1. Skill concentration
If your entire value is one narrow task inside one tool, you are concentrated.
Ask yourself:
- Which of my skills would still be useful at a completely different company?
- Do I understand the domain (health, finance, law, education), or only the task?
- Can I use AI tools to do my work faster, and explain how?
Practical move: pair your core skill with one adjacent skill. An annotator who learns evaluation design. A tester who learns basic automation with AI tools. A content writer who learns analytics.
2. Employer and client concentration
Most salaried people have one employer. That is normal. The risk is when that employer is also your only source of opportunities, contacts and learning.
Ask yourself:
- If I lost access to my work laptop today, could I reach ten people who know my work?
- Do I know which clients or projects pay for my team?
- Do I have a second channel for opportunities, such as a community, a referral network or occasional freelance work?
Practical move: find out, politely, what your team's work depends on. And keep your professional contacts somewhere you own, not only in company chat.
3. Proof-of-work concentration
The Innodata accounts include people who could not even download their payslips once access was cut. If all evidence of your work lives inside company systems, it can disappear in an afternoon.
Ask yourself:
- Can a stranger see anything I have built, improved or explained?
- Do I keep my own copies of payslips, offer letters and appraisal records?
Practical move: build a one-page portfolio. Two or three projects, described without confidential details: the problem, what you did, what changed. Small AI-assisted prototypes count. They show initiative and they belong to you.
A 30-minute weekend checklist
- Download and save your last six payslips, offer letter and latest appraisal to personal storage.
- Write a one-line description of your value with no company name.
- List the three skills that would travel to another company. Circle the weakest.
- Note one adjacent skill to learn in the next 90 days.
- Draft your first portfolio entry.
None of this needs you to quit, panic or predict the future of AI. It just makes you less of a one-client business.
A note of care
If you were affected by the Innodata layoffs, I am sorry. A sudden exit like this is hard, especially with family responsibilities. Request your documents and settlement details in writing, and lean on the people around you. This article is not about blame. It is about making the next chapter sturdier.
If you want structured help building skills and proof of work that travel across employers, take a look at our programs at The Arthi AI Collective.