The Model Is Breaking

The signal this week is not one thing. It is a pattern. The Indian IT outsourcing model - thirty years in the making, the backbone of American enterprise technology - is fracturing. The Saudi Cabinet has declared 2026 the Year of AI and built the world's largest government data centre to prove it. And MIT researchers have put a number on the pace of AI capability growth that should make every workforce planner reconsider their assumptions. Three signals. One direction.

Signal 1: India's Silent Layoff Wave

WARN notices filed. Contractors invisible. The count is wrong.

In 2024, Indian IT firms filed four WARN notices in the United States. In 2025, only Genpact filed one. In the first quarter of 2026 alone, Infosys, HCL Technologies, and Hinduja Global Services have already filed multiple notices - across Florida, Texas, and Pennsylvania. WARN notices are required by US law for large layoffs. They are the most reliable public record of what is actually happening on the ground. The acceleration is stark.

The mechanism matters as much as the numbers. The old model worked like this: a US bank or healthcare system signed a multi-billion dollar, multi-year outsourcing contract. The Indian IT firm deployed hundreds of engineers on US soil - rebadging employees from the client company - to manage the project over its duration. AI is now shrinking the headcount those projects require. Productivity gains of up to 30% mean fewer people are needed for the same work. When the project winds down, the rebadged workers cannot be redeployed elsewhere because the same productivity gains have reduced demand across the portfolio.

What makes this signal particularly acute for reAImagine.work's audience is what the WARN notices do not capture. Indian nationals employed as contractors - not direct employees - do not appear in layoff statistics. They cannot claim unemployment benefits. They disappear from the data while remaining entirely present in the human cost. The India IT workforce story is being systematically undercounted by the metrics designed to measure it.

Why this matters for your work: If your organisation has large transformation programmes managed by Indian IT service providers, model what a 25-30% AI productivity gain means for the headcount those contracts require. The renegotiation conversation is coming whether you initiate it or not.

Signal 2: Saudi Arabia Declares 2026 the Year of AI

Not symbolism. 480 megawatts of compute and a mandatory university curriculum.

In March 2026, the Saudi Cabinet officially designated 2026 as the Year of Artificial Intelligence under the patronage of Crown Prince Mohammed bin Salman. The Kingdom has inaugurated Hexagon - the world's largest government data centre at 480 megawatts capacity. The Shaheen III supercomputer is live. A National Data Lake integrating over 430 government systems is operational. Saudi Arabia now ranks 14th globally in the 2025 Global AI Index and first in the Arab world for advanced AI model development according to Stanford University indicators.

For workforce leaders in the GCC, the signal is not the infrastructure - it is the human capital agenda behind it. Saudi Arabia has made AI curriculum mandatory at university level. The fourth Global AI Summit is scheduled for September 2026. The Kingdom has become the first Arab nation to join the Global Partnership on AI.

The tension this creates for Saudization is significant and largely undiscussed. The Year of AI accelerates automation in exactly the sectors - financial services, logistics, government services - where nationalisation targets are being applied. The workforce strategy question is not whether Saudi nationals will fill these roles. It is whether the roles they are being placed into will exist in their current form by the time the placement programmes complete.

Why this matters for your work: If you are a GCC organisation with active Emiratisation or Saudization commitments, map the AI exposure level of the roles your nationals are currently being placed into. Compliance reporting counts placements. It does not count the automation trajectory of the roles being counted.

Signal 3: MIT Says AI Completes 65% of Text Tasks - Up From 50% Eighteen Months Ago

The pace data that invalidates every three-year workforce plan.

A new MIT study - measuring real-world workplace performance rather than benchmark scores - found that AI models can now complete 65% of text-based tasks at a minimally acceptable level without human edits. That figure was 50% in 2024. At the current pace of improvement, MIT projects AI will handle 80-95% of text-based tasks by 2029. The study evaluated 11,500 tasks from the US Labor Department database across more than 40 AI models, with workers in each field evaluating whether the outputs were usable.

The framing that matters is not the 65% figure in isolation. It is the rate of change. From 50% to 65% in eighteen months is a pace that invalidates any workforce planning assumption built on a three-to-five year horizon. Most reskilling programmes are designed for a world where organisations have time to identify which roles are at risk, develop training pathways, and transition workers over multiple cycles. The MIT data suggests the window for that approach is closing faster than the programmes can run.

The nuance the MIT study itself emphasises - and that most coverage misses - is that "minimally acceptable" is not "excellent." AI completing 65% of text-based tasks at a passing grade does not mean 65% of those tasks are fully automated. It means the human role in those tasks shifts from execution to quality control, judgment, and the 35% that AI cannot yet do. That is a different kind of displacement - not elimination but radical redefinition - and it is harder to plan for than simple job loss.

Why this matters for your work: Run the MIT framework against your own organisation's role inventory. What percentage of the tasks in your highest-headcount roles are text-based? Of those, what percentage require "minimally acceptable" output versus expert-level judgment? The gap between those two numbers is your actual automation exposure.

Three signals, one pattern: the model that has governed global workforce deployment for thirty years is being repriced by AI at every node simultaneously. The question is not whether the transition is happening. It is whether your organisation has the eighteen months the MIT data suggests you have left.