The Fallback Economy: The Graduate on the Delivery Bike

The scale of the fallback

NITI Aayog's 2022 report, India's first attempt to size a workforce that had grown largely outside official labour statistics, counted 7.7 million gig and platform workers in 2020-21. By the government's own account in the Economic Survey 2025-26, that number had grown to about 12 million by FY25, and official projections put it at 23.5 million by 2029-30, or roughly 6.7 percent of India's non-agricultural workforce. Coverage of the same projections in Indian business press points to about 2 million new gig roles opening in 2026 alone, most of them driven by the country's fast-expanding quick-commerce sector.

The growth is real and it is fast, close to doubling in under five years by the government's own count, with a further near-doubling projected inside the next four. The size of the number is only part of the story. What filled it matters more. NITI Aayog's original 2022 framing treated gig work chiefly as opportunity, flexible income for people who chose it around study, caregiving, or other work. By 2025-26, the government's own language had shifted: the Economic Survey describes workers who are pushed into gig work by weak formal-sector demand and a mismatch between what graduates are trained for and what employers are hiring for, a marked change in how the same institution talks about the same workforce inside three years.

A workforce this size is also large enough to be a genuine labour-market shock absorber rather than a marginal category. Six or seven out of every hundred non-agricultural workers being routed through gig platforms by 2029-30, on the government's own projection, is not a niche outcome; it is a structural feature of how India now employs people who cannot find a formal job. That scale is what makes the composition of the workforce, who is actually taking these jobs, the more important question than the headcount alone.

What changed between the 2022 report and the 2025-26 survey was not the definition of gig work. It was who is doing it, and why they say they joined.

The degree on the delivery bike

The clearest evidence of who is filling these roles sits in the mismatch between India's graduate output and its formal job market. The Economic Survey 2024-25 put a number on that mismatch directly: only 8.25 percent of India's graduates hold a job that matches their qualification. More than half of all graduates, and about 40 percent of postgraduates, are underemployed, working below the level their degree would suggest they should reach.

Independent surveys of the delivery workforce line up with that gap rather than contradicting it. A 2024 survey by the delivery-logistics platform Borzo found about 29 percent of its delivery workers held a formal degree. A 2023 study by the National Council of Applied Economic Research put the graduate share of India's delivery workforce at just over 30 percent. Neither figure describes a workforce made up mainly of school leavers picking up casual income between other plans; both describe a meaningful and measured share of India's graduate population working delivery routes as their primary income.

The clearest single data point comes from a March 2025 study by the research organisation IDinsight, which surveyed 2,547 delivery drivers directly rather than relying on a platform's own self-reported numbers. Forty-five percent had previously held a formal full-time job before joining a platform. Fifty-seven percent said they joined specifically to meet household financial needs. Only 6 percent cited flexibility, the reason gig work is usually marketed on to prospective workers and investors alike. Read together, these three figures describe less a lifestyle choice than a landing spot: a formal job lost or never secured, and a platform that was hiring at scale and without a formal interview process to screen a graduate degree out.

None of this means every delivery worker in India is an underemployed graduate; the workforce is large and mixed, and school leavers, students, and workers without a degree make up a real share of it too. What the surveys establish is that the graduate share is neither trivial nor a rounding error. It is large enough, and consistent enough across three separate studies using three separate methods, to count as a defining feature of who India's gig economy now employs.

The employer of last resort

The platforms absorbing this workforce are large by any national count. Zomato's parent company, Eternal, reported about 4.73 lakh, roughly 473,000, active delivery partners in FY25, a figure that turns over at a very high monthly rate as workers join and leave. Swiggy and Zomato together run more than 300,000 riders between them, and Rapido, the ride-hailing and delivery platform, reports about 2 million active earners across its network. A small number of platforms now account for a meaningful share of India's entire measured gig workforce.

What that work actually pays has become its own point of dispute between platforms and researchers. The IDinsight study put a full-time delivery driver's net pay, after expenses, at about Rs 18,761 a month, working out to roughly Rs 75 an hour across a 62-hour week. Base per-order payouts have been falling on some routes, from about Rs 40 to 45 a delivery down to Rs 15 to 20. The Economic Survey 2025-26 reports that about 40 percent of gig workers earn below Rs 15,000 a month, and the Borzo survey found 78 percent earning under Rs 2.5 lakh a year, close to Rs 20,800 a month on average.

Working conditions carry a similarly mixed record when measured independently rather than described by the platforms themselves. The Fairwork India 2024 ratings, which score platforms on pay, conditions, contracts, management, and worker representation, gave no platform a score above 6 out of 10. Delivery partners staged nationwide strikes in December 2025 over falling per-order payouts and account deactivations that workers said came with little explanation or appeal.

None of this describes an industry in crisis by revenue; quick commerce, the segment doing most of the recent hiring, is one of India's fastest-growing retail categories, and the platforms running it are scaling their store networks aggressively. It describes an industry whose growth and its workers' economics are not moving in the same direction, and where the size of the workforce, its degree of formal-job experience, and its earnings ceiling are each measured separately but tell a coherent, converging story.

The algorithm cuts both ways

The same technology reshaping India's white-collar hiring is also the technology managing the workers who land in gig work after losing, or never finding, a white-collar job. The Economic Survey 2025-26 names this directly, flagging what it calls algorithm-driven work allocation as a defining feature of the gig economy, alongside income volatility and the absence of a formal safety net for workers who fall under it. Workers can be logged out or deactivated by an automated decision with limited recourse, and the ten-minute delivery model that quick commerce is built on compounds the pressure on them: earnings depend on a pace the algorithm sets, not one the worker negotiates.

The scale of that model is worth stating plainly, because it is the actual mechanism behind the numbers above, not a figure of speech. India's quick-commerce market reached about Rs 64,000 crore in 2025-26, running roughly 1,900 dark stores nationally, a network industry estimates project will pass 5,000 in the coming years. Every one of those stores runs on the same allocation logic: an automated system assigns the order, times the delivery window, and scores the worker against both, with pay and future access to shifts following the score rather than a human manager's judgement.

At the other end of the pipeline, the entry rung graduates once climbed into formal employment has been thinning for a different, related reason. India lost about 100,000 IT jobs in 2025, and hiring for entry-level, or "fresher", roles in IT fell by roughly 80 percent over three years, from about 600,000 openings in FY22 to about 120,000 in FY25. The World Bank noted in October 2025 that, unlike earlier waves of automation which mainly displaced routine manual and clerical work, AI is now capable of displacing non-routine white-collar service work, precisely the category of job that large-scale IT hiring used to offer new graduates as their first formal role.

Put the two halves of this section next to each other and the shape becomes clear. AI is not one technology doing one thing to India's labour market. It is thinning the formal entry rung that graduates used to climb, and it is simultaneously running the allocation system that manages the gig work many of them fall into once that rung is gone. The same wave shows up on both ends of the same worker's path.

What connects the two rungs

No single dataset tracks an individual worker from a lost IT job to a delivery-app sign-up; that specific link is a reasonable inference from adjacent data, not a measured pipeline. But three separate facts sit close enough together in time and direction to support a plausible reading. The white-collar entry rung is thinning, with fresher IT hiring down by roughly four-fifths in three years. The gig workforce is growing at a pace close to the rate the formal entry rung is shrinking, adding an estimated 2 million roles in 2026 alone. And the closest direct measurement available, IDinsight's finding that 45 percent of delivery drivers had previously held a formal full-time job, shows the traffic between formal and gig work running in exactly that direction, even without identifying AI as the specific cause in each individual case.

The government's own change of language captures the same shift without needing a single causal study to state it outright. NITI Aayog's 2022 report described a workforce broadly choosing flexible income. The Economic Survey 2025-26 describes a workforce pushed into gig work by weak demand for formal labour and a mismatch between what graduates are trained for and what the market is willing to pay for. Between those two documents sit three years in which fresher hiring collapsed, graduate overqualification became a measured and published fact rather than an impression, and the gig workforce absorbed a rising share of whoever was left without a formal option.

This is not a story of one villain technology and one victim workforce. It is closer to a redistribution: work that used to be organised through a formal hiring process, with a manager, a salary band, and a career ladder, is increasingly organised through an app, an algorithm, and a per-order rate. The graduate on the delivery bike did not choose that redistribution. The data suggests they are, in growing numbers, living inside it.

The limits of this reading

The "riders out-earning engineers" story that circulates online is real only at the very top of the distribution. A small share of delivery workers, those putting in very long hours on the highest-value routes, can clear Rs 40,000 to 50,000 a month, comparable to an early-career formal salary in several white-collar fields. That is not the median. IDinsight's measured figure for a full-time driver, after expenses, is about Rs 18,761 a month on a 62-hour week, and the Economic Survey puts 40 percent of the entire gig workforce below Rs 15,000 a month. The exception is the one that gets shared online; the median is the one that describes the actual economics of the work.

The causal claim at the centre of this piece, that AI-driven thinning of white-collar hiring is feeding India's gig economy, is best read as a plausible and well-supported inference rather than a demonstrated fact. No public, nationally representative survey has tracked laid-off white-collar workers directly into gig platforms and asked them why they joined. The closest proxy, IDinsight's 45 percent figure, confirms that a large share of delivery drivers previously held formal jobs; it does not establish that AI, specifically, cost them that job, as opposed to a hiring slowdown, a company closure, a personal circumstance, or an unrelated reason. This distinction matters and is kept deliberately visible rather than smoothed over.

A further note applies to the numbers themselves. The 23.5 million figure for 2029-30 is a government projection, not a count, and should be read as a scenario rather than a certainty. Several of the earnings and workforce figures used here, including the platform-level rider counts and some of the IT hiring figures, are secondary-sourced, reported in business press rather than published in full detail by the original survey house, and are tiered accordingly. Gig work is also not a uniform bad outcome for everyone inside it: for some workers it is genuine, chosen, flexible income around study or other commitments, and the Fairwork ratings and the December 2025 strike coverage describe a labour-conditions and pay problem specific to the current terms of the work, not a blanket indictment of gig work as a category.