Career Choices After Class 12 in the AI Era: What the Evidence Supports

Solution via Claude with ref to news article Zoho founder Sridhar Vembu warns of looming jobs crisis, says 'problem is about to get worse'

 Career Choices After Class 12 in the AI Era: What the Evidence Supports

India, November 2026. Based on a fresh review of news reports, official releases and industry data.

How to read the figures

• [V] Verified in a news report or official release that I opened.

• [S] Secondhand: a news site citing a study.

• [I] Institute or vendor figure: from colleges, course sellers, job-prep sites or software vendors with a commercial interest. Directional only.

• [R] My ( Claude's) reasoning (as no credible source found in required resources).

1. The premise, checked

• Vembu's recent position [V]: On 29 October 2026 he said any job done on a computer now faces the risk that one person using AI does the work of ten. He added that this does not mean 90 percent of such workers lose work, because businesses that could not afford software engineers or digital marketers can afford them now. He called it a painful rebalancing and said the only course is to accept it and prepare.

• Zoho itself [V]: Per the same report, Zoho cut its marketing budget and told staff not to expect the usual pay increases this year.

• On education debt [V]: In August 2025 he urged students and parents to be cautious about heavy education loans, citing weak IT hiring.

• His earlier hedge [V]: In February 2026 he said he was still early in understanding AI's effect, and called a post arguing white-collar jobs will not vanish soon a strong argument.

• Entrepreneurship advice: I found it only in a short aggregator summary, not in a full source. Treat it as unconfirmed.

• [R] His position is large disruption, not collapse. The draft's framing is fair on disruption. It overreaches if it implies mass white-collar disappearance.

2. Decision rules for a Class 12 student

• Rule 1: Prefer work with a licence, physical presence, liability or a trust relationship. These resist automation.

• Rule 2: Check whether the AI-skilled version of the job pays more than the unskilled one. If yes, learn the AI-skilled version.

• Rule 3: Build proof of work by age 19 or 20: internships, a portfolio, certifications backed by real projects.

• Rule 4: Discount fields that depend on one government scheme or on course-seller forecasts.

• Rule 5: Avoid heavy education debt (or expenses) for degrees with weak placement, in line with Vembu's caution.

• Rule 6: Treat AI literacy as a baseline in every field, not a separate career.

• Rule 7: Treat entrepreneurship as a sequence: skill, then paid work, then a side venture, then a business.

3. What the entry-level market shows

• Cuts [S]: Nomura reportedly found 55 percent of Indian IT firms cut entry-level hiring against 14 percent for senior roles. A 20 to 25 percent entry-level cut is attributed to an EY report.

• Rebound signals [V]: LinkedIn reports entry-level hiring in India rose 168 percent between 2023 and 2025, driven by AI roles, internships and hiring beyond the metros. Naukri JobSpeak showed fresher hiring up 17 percent in February 2026, led by non-IT: BPO/ITES up 48 percent, insurance up 42 percent. India's IT-BPM sector is projected to hire about 2.1 lakh freshers in FY27, a recovery from the decline.

• Split by skill [V/S]: TCS reportedly said 60 percent of its FY26 fresher hires were AI-skilled, up from 10 to 15 percent three years earlier. Fresher pay at the same large recruiters ranges from about ₹3.4 lakh (TCS Ninja, ₹3.36 lakh) to ₹21 lakh (Infosys specialised roles) [I]. A Nasscom-Indeed study is cited for a 30 to 40 percent premium for AI-skilled professionals [S].

• Hollowing pattern [V]: Premium and senior hiring grows faster than bulk entry-level and commodity software roles.

• [R] The market is not shut to freshers. It is splitting. A generic entry-level worker in any computer-based role is exposed. The same person with proof of AI-assisted, role-specific skill is not.

4. The self-employment trap

• Already the majority [V]: India's PLFS 2025 shows 56.2 percent of workers self-employed and 23.6 percent in regular salaried work.

• Earnings gap [S]: ILO-IHD data cited in Outlook Business put average self-employed earnings at ₹10,201 a month against ₹15,177 for regular wage workers in 2022. State of Working India is cited for a decline in real self-employment earnings after 2019.

• Salary benchmark [S]: PLFS 2025 average regular-salaried earnings are reported at ₹24,217 a month for men and ₹18,353 for women.

• Youth unemployment [V]: PLFS 2025 puts urban youth unemployment at 13.6 percent and rural youth at 8.3 percent.

• Distress vs dynamism [S]: Analysts note that much self-employment in India is own-account or unpaid family work, a fallback rather than enterprise.

• [R] 'More self-employment' is not the same as 'more opportunity.' The entrepreneurial path works when it starts from a skill customers already pay for, not from the wish to avoid a job.

5. Field by field

Allied healthcare (lab, radiology, OT, dialysis)

• Demand [V]: Union Budget 2026-27 backs adding 1 lakh allied health professionals over five years across 10 disciplines. India has about 12 to 13 lakh today.

• Gap claim [I]: 30 to 40 percent unmet demand, and a shortage of 80,000+ OT technologists. Both come from institutions that sell these courses.

• Pay [I]: Fresher ₹1.5 to 4.5 lakh across most sources; experienced ₹4 to 8 lakh. Some course sellers quote ₹2.5 to 8 lakh for imaging.

• [R] Hands-on, licensed, demand-backed. Low ceiling early. Strongest as a stepping stone to a degree, specialisation, a Gulf posting or a diagnostics or support-services business.

Solar and renewables

• Scale [V]: PM Surya Ghar has crossed 60 lakh households and about 18 GW of rooftop capacity, with 37,000+ registered vendors and 2.5 lakh people trained. The skilling plan targets about 3 lakh workers, including 1 lakh solar PV technicians.

• Bottleneck [V]: pv magazine India (October 2026) calls skilled manpower the next constraint.

• Counter-evidence [V/S]: NISE's 2024-25 annual report shows 30,386 of 60,685 trained Suryamitras placed, about 50 percent. A separate report says trained technicians struggle to find jobs because rooftop work lacks annual maintenance contracts.

• Policy risk [S]: Scheme guidelines describe implementation until FY 2026-27. Check what follows.

• [R] Real demand, but subsidy-linked. Better as an installer or maintenance business with a vendor tie-up than as a salaried job. Training alone does not guarantee work.

EV service and charging

• Pay [I]: EV service technician ₹2 to 5 lakh at entry; charging technician ₹2 to 4 lakh; public submissions show ₹2 to 6 lakh overall.

• Forecast spread [I/S]: Projections range from about 1 lakh service jobs by 2030 to 1.7 million by FY2030, and 3.7 to 5.6 million by 2047. A spread that wide means no forecast is reliable.

• [Input] ₹4 to 12 lakh is unsupported for technicians. Higher pay sits in battery and power-electronics engineering.

• [R] Good physical-skill niche, needs certification, pays modestly early. Franchise or fleet-service ownership is the upside path.

Cybersecurity

• Pay [I]: Fresher ₹3.5 to 7 lakh in most sources. Without verified experience, offers cluster at ₹3.5 to 6 lakh. With a structured internship, ₹6 to 10 lakh.

• Gap claims [I]: Figures like 80,000 experts against 1.5 million needed circulate widely, but they come from training sellers and I could not trace them to a primary study.

• [R] Demand is real, but the entry barrier is the 'fresher tag,' not certificates. Level-1 SOC monitoring is also the most automatable layer. Aim for cloud security or governance and compliance work.

Accounting, GST and compliance

• [I] Nearly all sources are accounting-software vendors or exam-prep sites. They agree that data entry, reconciliation and GST filing are being automated while judgment and advisory remain.

• [R] Licensed work (CA, CMA, ACCA) keeps its moat. Clerical bookkeeping loses entry-level seats. Pure 'GST filing' as a business is shrinking in value.

Software, data and AI roles

• [V/S] See Section 2. Demand is shifting to AI-skilled hires at a pay premium, while generic entry roles are cut.

• [R] Coding is no longer a moat on its own. Pair it with a domain: finance, healthcare, logistics, retail.

Sales, digital marketing, logistics, UI/UX, industrial automation

• Gap: I found no solid independent data for these. Everything below is reasoning.

• [R] Sales and business development rest on trust and negotiation, which are hard to automate. Digital marketing has low entry barriers and heavy competition; Vembu's own point is that AI makes small-business marketing cheaper, which may expand demand but compress prices. Logistics and industrial automation are physical plus technical, but I have no figures to rank them.

6. Decision rules for a Class 12 student

• Rule 1: Prefer work with a licence, physical presence, liability or a trust relationship. These resist automation.

• Rule 2: Check whether the AI-skilled version of the job pays more than the unskilled one. If yes, learn the AI-skilled version.

• Rule 3: Build proof of work by age 19 or 20: internships, a portfolio, certifications backed by real projects.

• Rule 4: Discount fields that depend on one government scheme or on course-seller forecasts.

• Rule 5: Avoid heavy education debt (or expenses) for degrees with weak placement, in line with Vembu's caution.

• Rule 6: Treat AI literacy as a baseline in every field, not a separate career.

• Rule 7: Treat entrepreneurship as a sequence: skill, then paid work, then a side venture, then a business.

7. Pathways by stream

• Science: MBBS, BDS, nursing, pharmacy and allied health for licensed healthcare. Engineering in electronics, EV, robotics or computing, paired with a domain. Data and AI with a project portfolio.

• Commerce: CA, CMA, ACCA, finance, sales and business development. Add data analytics to any of them.

• Arts and humanities: Sales, marketing, content, law, design, policy and psychology. Add one hard skill: analytics, a language or a technical tool.

• Any stream: Vocational skills such as solar, EV, lab or imaging technology, plus digital marketing, UI/UX and cybersecurity, each with a work placement.

8. Bottom line

• The evidence favours one marketable skill with proof of work, sales or communication ability, AI literacy, and a delayed, customer-led move into entrepreneurship.

• Self-employment alone is not a strategy in India. Over half of workers already do it, often at lower earnings than salaried peers.