• Home
  • NE100
  • Features
  • Brand Voice
  • Innovation
  • Leadership
  • public policy
  • collection
  • Video
    • Current issue
    • Magazine issue undefined
Echelon logo
  • Features
  • Portfolio
  • Brand-voice
  • Innovation
  • Leadership
  • Public-policy
  • Collection
  • Videos
Sri Lanka Trains 17,000 Coders A Year For Jobs AI Is Erasing
Sri Lanka Trains 17,000 Coders A Year For Jobs AI Is Erasing
Sep 17, 2026 |

Sri Lanka Trains 17,000 Coders A Year For Jobs AI Is Erasing

Tech Has Overtaken Tea As An Export Earner. Now The Industry Says Most Graduates Aren’t Hireable, And Cuts Will Run To 2028

Sri Lanka’s technology industry, which overtook Ceylon tea as an export earner in 2024, is shedding jobs and will keep shedding them until about 2028, even as universities turn out a record 17,000 information-technology graduates a year that the industry says it mostly can’t hire.

Technology exports earned $1.64 billion in 2025 against $1.51 billion for tea, according to Export Development Board data, and the direct workforce has grown to about 155,000 from 144,000 three years ago. Universities produced roughly 3,000 IT graduates a decade ago. Only a few thousand of today’s 17,000 can start work without further training, said Sampath Jayasundara, Chairman at SLASSCOM (Sri Lanka Association of Software and Service Companies), an industry trade group.

“The majority is not industry-ready,” Jayasundara said in an interview. Artificial intelligence won’t replace people, he said, but “it will replace people without AI skills.”

SLASSCOM wants $5 billion in exports. Growth already underway carries the industry to $3 billion, Jayasundara said. The missing $2 billion was meant to arrive the way the earlier billions did, by hiring, and hiring is the assumption AI has broken. Clients adopting the technology themselves are asking why a bill should scale with the number of engineers, and the junior work that used to fill those desks is what AI does best. Sri Lanka spent a decade building a graduate pipeline to feed them. Jayasundara expects the next wave of demand around 2029, for skills few local institutions teach yet.

The Vanishing First Rung

By the old measure of success, a nearly sixfold rise in graduate output would count as a problem solved. It isn’t.

A familiar programming language and a degree from a government university used to guarantee a job. The question now, Jayasundara said, is not whether a graduate uses AI tools. “It’s about whether you can build that skill. Are you capable of building something with AI?”

The industry has a name for what it wants: the builder. Jayasundara describes it as a return to how software work was done before it was split into narrow roles. “Builder means what we used to do 30 years back,” he said. “We used to go and meet the customer, understand their problems, develop the design, code, then quality assurance, deployment, and training.”

A graduate once reached that job over years, starting as a junior coder and gradually learning a client’s business. AI has removed the runway. What used to arrive with experience is now expected on day one.

Jayasundara has no precise count of how many of the 17,000 clear the bar. “Obviously, a few thousand are ready,” he said. “But the balance, you have to guide them, you have to coach them.” He calls the gap closeable, pointing to free courses, online platforms, and AI tools themselves. “If they can learn AI, they can find work.”

Cuts Run to 2028

The disruption began in 2024, Jayasundara said. “Now we are in the middle.” He expects further job losses over the next two years, and veterans are exposed alongside graduates. “No point in saying I have five years’ work experience,” he said. What counts is whether that engineer can learn the technologies the work now requires.

At Disrupt Asia 2026, Mani Kulasooriya, founder of Sysco LABS, framed the choice facing a services company as shrinking to three people doing what 10 once did, or keeping the 10 and having them do the work of 30. Jayasundara agreed and said the second describes his own firm.

Jayasundara has led hSenid Business Solutions, an HR software company, since 2007 and is the first enterprise software company listed on the Colombo Stock Exchange. It employs 371 people. Revenue grew from Rs1.65 billion in the year to March 2024 to Rs2.1 billion in the year to March 2026, while investing heavily in sales and product development. It’s an investment that appears to be paying off: hSenid swung to near-breakeven in the financial year ending March 2026.

He began using AI about two years ago to expand demand generation rather than cut staff, and has since added around 10 people, screening for how fast they learn rather than what they already know. “I don’t care whether they know AI now,” he said. “But after three months or so, if they don’t know, then that’s a problem.”

Not every company can make that choice, he acknowledged, particularly those under pressure to stay globally competitive. His own view puts the burden on the individual. “If I’m an AI-ready engineer, there’s no reason for a company to move me out.”

Paying for Outcomes

Sri Lankan firms have charged the way most outsourcing companies do: a client pays for a set number of engineers for a set number of hours, whatever gets delivered. Clients worldwide are now asking why they should keep paying by headcount when a smaller, AI-equipped team produces the same result. Some outsourcing firms have moved to outcome-based pricing. Others globally are pushing for longer contracts of five to ten years.

Local companies are making the same moves. “It’s a very early stage at the moment,” Jayasundara said, “but the company should be ready for that challenge.” SLASSCOM is working with members on alternative models, including the subscription pricing common in software-as-a-service. “You’ve got to educate companies who want to understand these models so that they can adapt,” he said.

Two Pillars

SLASSCOM’s answer rests on redesigning what universities teach and re-skilling the 155,000 people already employed. Using what Jayasundara calls an 80/20 rule, the association is identifying the roles that cover most of the workforce and writing a persona for each, defining how AI has changed the job and what it now demands. “We don’t want to keep it,” he said. “We want to make it open source.”

The association isn’t trying to do the teaching. “We can’t train 100,000 people,” Jayasundara said. Because redesigning a degree programme takes three to four years to reach students, SLASSCOM shares the personas with universities and encourages short, targeted modules instead. A companion programme embeds lecturers in private companies for two to three weeks. Member companies use the same personas for in-house training, alongside SLASSCOM Academy, whose 250-plus programmes have trained more than 19,000 professionals, according to the association’s figures.

The goal is to sell something more specific than cheap capacity. “We have to move our industry from a mere outsourcing destination to a high-value, talent creation, digital destination,” Jayasundara said. “If you build the workforce, we can go out and say: bring your GCC to Colombo.”

Hockey Stick

Asked where the industry lands in five years, Jayasundara reaches for a shape rather than a number. “I would say it is like a hockey stick” — the dip first, as pay models, headcounts, and hiring standards all reset at once, then the climb.

Company size will matter less than it did, he said. Large firms will grow by retraining internally, but “even a small company can produce larger outcomes” as AI lowers the cost of building software. “There’s a massive opportunity in front of us,” he said. “But it all depends on whether we make that change.”

Most Popular

Advertisement
© 2026 Echelon Media (Pvt)Ltd. All Rights Reserved.
  • Features
  • Portfolio
  • Brand Voice
  • Innovation
  • Leadership
  • Public Policy
  • collection
  • About Us
  • Contact Us
  • Privacy Policy