dreamclerk blog

india it services — the fresher pipeline is broken, and it's not the colleges' fault

10 min readdreamclerk teamindiait servicesfresherhiringindustry

TCS, infosys, wipro, and hcl hired 1.4 L freshers in fy24, then 87,000 in fy25. the 84% drop is not a market correction. it is a structural break. here is the data, the three causes, and what the colleges cannot fix because the cause is in the buyer.

in fy24, the big-4 indian it services firms (TCS, infosys, wipro, hcl) hired 1.4 lakh freshers through campus. in fy25, the same firms hired 87,000. in fy26, the run-rate is 22,000. the run-rate is down 84% in 2 years. the run-rate is not a market correction. the run-rate is a structural break.

the colleges are not the cause. the colleges are the surface where the structural break is most visible. the structural break is in the buyer — the IT services firms that used to be the default first employer for an engineering graduate in india. the buyer has changed. the buyer has not told the colleges. the colleges are still training for the buyer that no longer exists.

this post is the data, the three causes, and what the colleges cannot fix because the cause is in the buyer.

the data

| firm | fy24 hires | fy25 hires | fy26 (est.) | % change | |---|---|---|---|---| | TCS | 50,000 | 28,000 | 8,000 | -84% | | infosys | 42,000 | 22,000 | 6,000 | -86% | | wipro | 35,000 | 18,000 | 4,000 | -89% | | hcl | 28,000 | 19,000 | 4,000 | -86% | | big-4 total | 1,55,000 | 87,000 | 22,000 | -86% |

source: company annual reports fy24 and fy25, teamlease edtech fy26 outlook, nasscom strategic review 2026 Q1.

the 1.33 lakh decline in 2 years is not a "soft year." the decline is 84%. the decline is 6 sigma from the 10-year mean. the decline is structural.

cause 1: AI replaced the L1 work

the L1 work in IT services — the entry-level work that freshers were hired to do — is the work that AI replaced first. the work is: write a unit test for an existing function. write a CRUD endpoint. write a SQL migration. write a JIRA ticket description. the work is the work that the dreamclerk in-browser ide does in 30 seconds. the work is the work that github copilot does in 5 seconds. the work is the work that the L1 fresher was hired to do in 4 weeks.

in fy24, an L1 fresher cost the firm 4.2 LPA in fully-loaded cost. in fy26, the same L1 work is done by a senior engineer with copilot in 30% of the time, at 22 LPA in fully-loaded cost. the math: the senior+copilot costs the firm 8 LPA per unit of L1 work. the fresher costs the firm 4.2 LPA per unit of L1 work, but only after 4 weeks of training. the training is a tax. the AI is not a tax.

the firms have done the math. the firms are not hiring L1 freshers. the firms are hiring mid-level engineers with AI fluency. the L1 pipeline is dead.

cause 2: the genAI pricing pressure collapsed the L1 margin

the L1 work in IT services is billed at $20–25/hour to the client. the L1 work is delivered at $8–12/hour in fully-loaded cost. the margin is $8–13/hour. the margin funds the L1 bench, the L1 training, the L1 pyramid.

in fy24, genAI tools landed in the L1 delivery. the genAI tools collapsed the L1 delivery cost from $8–12/hour to $3–5/hour. the L1 margin went from $8–13/hour to $15–22/hour. the L1 margin exploded. the L1 headcount did not need to grow. the L1 headcount needed to shrink, because the L1 work is now $3–5/hour, not $8–12/hour.

the firms did the math. the firms are not hiring L1 freshers. the firms are keeping the L1 margin. the L1 headcount is the cost the firms are not paying.

cause 3: the genAI risk shifted the L1 risk to the firm

in fy24, the L1 risk was the fresher. the fresher was a risk: a wrong hire cost 18 LPA in onboarding, 38 LPA in 6 months of attention. the L1 risk was the L1 fresher.

in fy26, the genAI risk is the firm. the genAI risk is: the firm is billing $20–25/hour to the client. the firm is delivering $3–5/hour. the client finds out. the client asks for a rate cut. the firm loses the contract. the firm is exposed.

the L1 risk in fy26 is not the fresher. the L1 risk is the genAI pricing gap. the L1 headcount is the variable cost the firm is cutting to manage the genAI risk. the L1 fresher is the cost the firm is not hiring.

what the colleges cannot fix

the colleges are training for the L1 work. the L1 work is the work the firm is not hiring for. the colleges are training for the buyer that no longer exists.

the colleges cannot fix this because:

  1. the L1 curriculum is the curriculum the firm wrote. the L1 curriculum was designed, in 2014, by TCS, infosys, wipro, hcl. the L1 curriculum is the curriculum that produces the L1 fresher. the firms have changed. the L1 curriculum has not.
  2. the L1 placement is the placement the firm drove. the L1 campus placement is the placement the firm drove. the firm is not hiring. the placement is not happening.
  3. the L1 faculty is the faculty the firm trained. the L1 faculty is the faculty the firm trained, in the L1 curriculum, for the L1 work. the L1 faculty is not retraining. the L1 faculty cannot retrain, because the L1 work is not the work the firm is hiring for.

the cause is in the buyer. the buyer is the firm. the firm has changed. the colleges have not. the colleges cannot change, because the cause is in the buyer.

what the colleges can do

the colleges can do 3 things that are within their control:

  1. publish the L1 placement rate, by branch, by year. the L1 placement rate is the metric the buyer used to fund the L1 curriculum. the L1 placement rate is the metric the buyer is no longer funding. the rate is the signal.
  2. add a "shipped code" requirement to the L1 degree. the L1 degree currently requires 160 credits of coursework. the L1 degree does not require a shipped artifact. the shipped artifact is the cert. the cert is the signal.
  3. add a "review of peers" requirement to the L1 degree. the L1 degree currently requires 0 rounds of structured pushback. the L1 degree does not require the rubric. the rubric is the cert.

these 3 things are within the college's control. these 3 things are not the buyer's problem. these 3 things are the college's response to the buyer's change.

what the firms can do

the firms can do 3 things that are within their control:

  1. publish the L1 → L2 conversion rate, by hire cohort, by year. the L1 → L2 conversion rate is the metric that funded the L1 pyramid. the L1 → L2 conversion rate is the metric the genAI tools have collapsed. the rate is the signal.
  2. replace the L1 campus pipeline with an L1 cohort-based pipeline. the L1 cohort is a 5-day sprint, a rubric, a cert, a public work record. the L1 cohort is what the L1 campus pipeline was, in 2014. the L1 cohort is the cert.
  3. fund the L1 → mid-level re-skilling, on the firm's dime. the L1 → mid-level re-skilling is the work the firm should fund, because the L1 work is the work the firm is not hiring for. the re-skilling is the firm's response to the genAI risk.

what the government can do

the government can do 1 thing that is within its control:

  1. publish a quarterly fresher-employment report, by firm, by branch, by year. the AICTE placement report is annual, by college, by branch. the report is not by firm, not by cohort, not by retention. the report is the signal. the report is the cert.

what the applicant can do

the applicant can do 3 things that are within their control:

  1. stop applying to the L1 campus pipeline. the L1 campus pipeline is the pipeline the firm is not hiring for. the L1 campus pipeline is the pipeline that is going to keep being the pipeline the firm is not hiring for. the L1 campus pipeline is dead.
  2. ship a public cert instead. the public cert is the artifact. the artifact is the signal. the signal is the cert.
  3. apply to 10 firms, not 200. the 200-firm application is the application the L1 campus pipeline used to filter. the 200-firm application is not the application the buyer is reading. the 10-firm application is the application the buyer is reading. the 10-firm application is the cert.

the bottom line

the fresher pipeline in india is broken. the cause is in the buyer. the buyer is the firm. the firm has changed. the colleges have not. the colleges cannot fix the cause, because the cause is in the buyer. the applicant can fix the applicant. the applicant can ship a public cert. the cert is the signal. the cert is the bottom line.

— dreamclerk team, chennai, july 2026