Global exports of digitally delivered services hit $5.26 trillion in 2025, up 10 percent from a year earlier, according to World Trade Organization (WTO) data. Computer services alone reached $1.22 trillion, more than double their 2019 level, while overall commercial services exports climbed to $9.56 trillion.
The theory behind that boom, laid out in a new economic analysis of how artificial general intelligence, or AGI, the still-hypothetical point where machines match human performance across most cognitive work, will reorganize high-tech services trade, never names a single country. Two economies have the most riding on it anyway. India’s and the Philippines’ 2026 numbers already show that theory’s contradictions playing out in real payrolls.
The Trade Theory That Never Names a Country
The intellectual root of the argument is “trade in tasks,” a framework economists Gene Grossman and Esteban Rossi-Hansberg introduced in a 2008 paper titled “Trading Tasks: A Simple Theory of Offshoring.” Their insight was simple: once communication technology gets cheap enough, firms stop shipping finished services and start shipping individual tasks instead.
A job, in this view, is a bundle of tasks. Software testing, data analytics, chip design and remote equipment diagnostics do not require the person doing them to sit in the same building as the customer. Once a task becomes digitally deliverable and organizationally separable, where it gets done turns into a sourcing decision rather than a geographic accident.
High-tech service work breaks into four rough layers, from the physical hardware at the bottom to the customer relationships at the top:
- Foundation layer – semiconductors, servers, data centers, energy grids and telecom networks
- Platform layer – cloud infrastructure, databases, software tools and foundation models
- Production layer – coding, model training, chip design and data analysis
- Control layer – system integration, domain expertise, intellectual property and customer relationships
Tasks scatter across dozens of countries. Control over the platforms, data, standards and customers those tasks feed almost never does; it stays concentrated in a small number of firms and hubs. The essay that lays out this architecture works entirely at that level of abstraction, talking about firms and countries in the plural without landing on Bengaluru, Manila, or any other place where the theory has a payroll attached.
Research from the Organization for Economic Co-operation and Development (OECD) backs the bundling idea: firms rarely move a single task abroad. They relocate whole business functions, IT support, R&D, engineering, marketing, back-office work, because splitting things too finely raises coordination costs faster than it saves money. The OECD has spent much of the past year building sharper experimental estimates of digital trade flows, precisely because official statistics still struggle to see this kind of work crossing a border.
India’s $315 Billion Tech Sector Becomes the Test Case
Nasscom, the trade association for India’s technology and outsourcing industry, projects the country’s tech sector will cross $315 billion in revenue for the fiscal year ending March 2026, up 6.1 percent. Nasscom’s Strategic Review 2026 puts the business process management segment at $59 billion of that total.
AI-linked revenue inside that number is still small: an estimated $10 billion to $12 billion, reflecting deployments that have moved past pilots into paid, function-specific products. More than 60 percent of business process management firms now rank generative AI as a top investment priority for the next three years.
The shift shows up in call centers first. The share of contact-center revenue coming from plain voice work is projected to fall from roughly half to somewhere between 30 and 40 percent over that same stretch, as generative AI assist tools take over routine calls.
The headcount math is the tell. Nasscom expects a net addition of about 135,000 jobs this fiscal year, pushing sector employment to near 5.95 million, roughly 2 percent headcount growth stacked against 6.1 percent revenue growth. Providers are re-pricing contracts around outcomes and shared risk instead of billing by the person, which is exactly the mechanism that lets revenue outrun payroll.
Manila Bets $25 Million on the Same Pivot
The Philippines is running a smaller version of the same experiment. The country’s IT-BPM sector grew about 5 percent in 2025 to roughly 1.9 million workers, and the industry is targeting close to $42 billion in revenue and 1.97 million jobs by the end of 2026.
Getting there means climbing the same stack India is climbing. The IT and Business Process Association of the Philippines, known as IBPAP, has committed at least $25 million a year to retraining workers for higher-value analytical and judgment-based work, an explicit hedge against AI eating the routine voice and back-office tasks the industry was built on. The industry’s own framing for that shift is a workforce that stays human-led even as AI takes on more of the routine load.
| Metric | India (FY26) | Philippines (2026 target) |
|---|---|---|
| Sector revenue | $315 billion (tech overall) | $42 billion (IT-BPM) |
| Employment | ~5.95 million, +135,000 net | ~1.97 million target |
| AI-specific figure | $10 billion to $12 billion in AI-linked revenue | $25 million a year committed to AI upskilling |
| Growth rate | 6.1 percent projected | ~5 percent in 2025 |
| Structural shift | Voice revenue share falling from ~50% toward 30-40% | Voice BPO shifting toward knowledge process work |
The two countries are not measuring identical things in that AI-specific row. One figure is booked revenue, the other is budgeted spending. The direction is the same: both are paying to move workers up the stack before automation moves the floor out from under them.
Will AI Gut the Philippines’ Outsourcing Economy?
No single answer holds. Philippine outsourcing employment kept growing through 2025 and the industry’s own 2026 targets still call for expansion, but multiple trade outlets report those targets have already been trimmed once because of AI-driven productivity gains, and opinion inside the industry splits over how much further that trimming goes.
- IBPAP’s public agenda – treats 2026 as human-led and AI-powered, betting $25 million a year on retraining to protect headcount targets
- Trade reporting from Outsource Accelerator – shows the industry group has already cut its own growth targets once, citing AI-driven productivity gains
- BusinessMirror columnist Rizal Raoul Reyes – argues the sector remains fundamentally strong despite the AI threat, pointing to steady demand for voice and back-office contracts
Both camps are reading the same industry. One side counts the jobs still being added. The other counts how many fewer jobs each new dollar of revenue now requires.
Freshers Still Get Hired as the Job Itself Changes
Tata Consultancy Services (TCS), India’s largest IT services company and one of the country’s biggest private employers, plans to onboard roughly 42,000 entry-level hires this fiscal year. Infosys, the Bengaluru-based IT major, is targeting more than 20,000 fresh hires in 2026. Neither number looks like a company bracing for a hiring freeze.
What has changed is the work those recruits do. Junior developers now spend less time writing routine code, since AI tools already handle most of that, and more time in front of clients translating business problems into specifications. TCS has built a dedicated hiring track for AI-focused roles at the early and mid-career level rather than folding AI skills quietly into old job descriptions.
This is the mechanism the original essay describes in the abstract: AI substituting for some cognitive tasks inside a job while making others, judgment, client interaction, systems thinking, more valuable. The occupation survives. Its content does not.
From Cheap Labor to Owning the Playbook
Bain & Company’s 2023 survey of more than 500 senior executives found 60 percent of companies planned to increase engineering and R&D outsourcing over the following three years. Seventy-three percent named industry or technology expertise as the leading factor in picking a supplier, against 59 percent who cited cost.
That gap is the whole story compressed into two numbers. Cost still matters. It has stopped being enough on its own. The upgrading path runs from executing standardized tasks, to managing an entire business function, to owning product modules outright, and only that last step produces the kind of scalable, bargaining-power-rich asset that survives a downturn or a new AI model release.
Participation in a value chain has never guaranteed a share of its value. What decides that is who defines the architecture, who owns the intellectual property and who holds the customer relationship. Performing a task earns a fee. Owning the software, the patent or the customer earns a multiple.
Governance Rules Become the Next Trade Border
Data-localization rules, privacy law, export controls and clashing AI-governance regimes now do the work a customs post used to do. In goods trade, the border shows up at a port. In task trade, it shows up inside a cloud contract or a technical standard deciding whose engineers can touch which dataset.
Trade-in-value-added data compiled by the WTO and World Bank, last updated for 2018, already showed services embedded in 50 percent of world trade once measured by value added rather than by what crosses a border. Services value added made up 31 percent of manufacturing exports in OECD economies and 29 percent in non-OECD economies. Manufacturing competitiveness now runs through software, data analytics and after-sales service almost as much as through the factory floor.
Nasscom’s own strategic review already assumes outcome-based, risk-sharing contracts keep replacing headcount billing through the rest of the decade. India and the Philippines are not waiting for AGI to arrive before restructuring around it.
Frequently Asked Questions
Is Artificial General Intelligence Here Yet?
No. AGI describes a hypothetical future point where machines match human performance across most cognitive tasks, and no research lab or government has agreed on when, or whether, that threshold gets crossed. The shifts already visible in India’s and the Philippines’ outsourcing sectors are being driven by current generative AI tools, not AGI itself; the essay treats AGI as the horizon that will accelerate a reorganization already under way.
What Is Knowledge Process Outsourcing, and How Is It Different from BPO?
Business process outsourcing (BPO) covers routine, rules-based work such as call-center support and data entry. Knowledge process outsourcing (KPO) covers higher-value analytical and judgment-heavy work such as research, legal analysis, engineering and specialized data science. The Philippine IT-BPM industry is explicitly steering investment toward KPO as generative AI automates the routine voice work that BPO built its scale on.
Why Is India’s Tech Sector Headcount Growing Slower Than Its Revenue?
Providers are shifting client contracts from per-person, per-hour billing toward outcome-based and risk-sharing pricing, so a project can generate more revenue without adding a proportional number of billable staff. Nasscom’s FY26 projections show sector revenue rising 6.1 percent against roughly 2 percent headcount growth, a gap industry executives attribute directly to AI-assisted productivity.
How Does the WTO Track Services Trade That Never Crosses a Physical Border?
The WTO and the OECD jointly maintain a trade-in-value-added database that traces how much of a country’s goods exports actually consists of imported services, and in February 2025 the two organizations expanded that dataset to cover more than 200 economies. It remains the main tool for seeing services trade that standard customs data misses entirely.
Which Skills Are Becoming More Valuable Inside Outsourcing Firms as AI Spreads?
Client-facing judgment, systems integration, domain expertise and accountability for outcomes are all commanding a premium, according to both the Bain survey and Nasscom’s sector data. Firms are also creating new roles around supervising AI output quality and prompt design, positions that barely existed inside outsourcing organizations three years ago.
Could AI Make Offshoring Disappear Entirely?
Unlikely. AI cuts both ways: it lowers the cost of coordinating work across borders through translation and monitoring tools, while simultaneously automating away many of the routine tasks that were easiest to send offshore in the first place. The likely outcome is fewer routine contracts, but larger, more complex and more valuable ones for the workers and firms that remain.
