On the first day of spring in February 2026, Anthropic in San Francisco quietly issued a press release, launching an enterprise-grade AI tool.

Thousands of miles away in Mumbai, the Nifty IT index's K-line plummeted like a kite with a broken string, crashing nearly 6% that day. It was the most devastating single-day decline Indian investors had witnessed since the March 2020 pandemic-induced market meltdown.

Three months later, OpenAI announced an investment of over $4 billion to assemble a massive enterprise AI deployment team. IT stocks in Mumbai promptly fell another 3.7%.

Over the past year, every cutting-edge release by Silicon Valley titans has triggered a precision-guided earthquake in Indian stock markets.

In this market once known as a "safe haven for foreign capital," money is leaving with unparalleled resolve. In the first half of 2026 alone, over $23 billion in foreign capital fled, driving the foreign ownership ratio down to 14.7%—a level not seen since the bleak days of fourteen years prior.

The Nifty IT index, hailed as India's version of the "Hang Seng Tech," has been in a persistent downtrend for 18 months, accumulating a 49% retracement—effectively halving in value. The ten largest IT giants have silently evaporated over ₹19 trillion from their balance sheets.

This sum could nearly cover 40% of India's entire national fiscal budget for last year.

But there are never collapses without cause.

For three decades, India thrived on answering global clients' calls and fixing low-level code, fully capitalizing on its demographic dividend while inadvertently positioning itself against technological progress.

When the price per token became cheaper than human labor along the Ganges, the outsourcing assembly line that once sustained countless Mumbai middle-class families suddenly lost all meaning before the cold calculus of computing power.

India, the world's most carbon-based human resource-abundant nation, not only failed to capture the incremental dividends of the silicon age but became the first casualty harvested by AI's scythe.

/ 01 /

The End of a Two-Decade "Labor Arbitrage Game"

To understand India's current pain, one must first grasp how it won in the past. The rise of India's IT outsourcing industry ultimately owes thanks to a bug named the "Y2K problem."

In 1999, legacy code in Western financial, aviation, and power systems using two-digit year records faced systemic collapse. Western enterprises urgently needed masses of programmers to troubleshoot (troubleshoot) and modify these antiquated codes. The work required little technical sophistication but involved backbreaking volume—digital-era "brick-laying."

Indians keenly sensed the opportunity. Leveraging three trump cards—"English proficiency, low wages, and ability to burn the midnight oil"—a cohort of Indian IT firms including TCS, Infosys, and Wipro rapidly ascended to become the "world's back office."

According to a 2025 report by India's National Association of Software and Service Companies (NASSCOM), the country's outsourcing industry had ballooned to a staggering $280 billion, directly employing 5.67 million IT engineers.

When accounting for each engineer's family and the ancillary dining, logistics, and property sectors revolving around them, nearly 25 million Indian middle-class livelihoods depend on this ecosystem.

Critically, this represents India's only large-scale foreign exchange-earning pillar industry, with IT services and BPO exports combined accounting for nearly a quarter of total Indian goods and services exports.

But when dissecting India's outsourcing business model, one finds extreme simplicity bordering on fragility: charging per headcount, billing by the hour.

An American programmer commands $150,000 annually, while an Indian engineer earns $15,000-$20,000; an American customer service rep makes $40,000, versus just $6,000 for their Indian counterpart.

Indian outsourcing firms undercut Western quotes by massive margins, then distributed wages domestically at local rates—profiting from this labor cost differential.

It was a perfect arbitrage game that Indians played for two decades, reaping enormous profits. Until AI arrived and flipped the table.

A 2025 study by scholars from Carnegie Mellon and Stanford delivered the fatal blow: AI agents complete tasks 88.3% faster than humans. Meanwhile, the 2025 median salary for engineering and data roles in India stood at $22,000, while AI programming tools cost merely hundreds to thousands of dollars annually.

When an indefatigable AI requiring no social security writes code 88% faster at a fraction of your cost, the very foundation of the "world's back office" crumbles instantaneously.

The chill has reached every Indian programmer's desk.

On April 29, 2026, global IT services giant Cognizant formally launched its transformation plan codenamed "Project Leap," setting aside $200-$270 million for severance. While specific numbers weren't disclosed, media reports claim 12,000-15,000 global layoffs, with the vast majority in India.

This was no isolated incident. U.S. proptech firm Opendoor shuttered all its Indian offices in Chennai and Bangalore; French pharmaceutical giant Sanofi handed over procurement order audits—previously outsourced to India—to SAP's AI agents.

The impact also manifests in financial reports. Industry leader TCS saw FY26 dollar-denominated revenue fall to $30 billion, down 0.5% year-on-year at constant exchange rates—its first annual revenue decline in years. Wipro's annual revenue stood at just $10.5 billion, down 1.6% year-on-year at constant exchange rates, virtually stagnant.

Even the most resilient Infosys, while breaking through $20 billion in revenue for the first time, grew at just 3.1% at constant exchange rates—far below its 13.7% compound annual growth rate over the past decade.

Layoffs are visibly accelerating. Global tech layoffs reached approximately 245,000 in 2025, with India contributing 19,000—second only to the U.S. Considering India accounts for far less than 7% of global tech employment, this represents 7.8% of worldwide tech layoffs.

More alarmingly, the trend has reversed. In FY25, India's top five IT firms still net added 12,718 employees; by FY26, they collectively shed 6,981 jobs. TCS alone cut over 23,000 positions, reducing its workforce from a peak of 614,000 to below 580,000—its largest net reduction since the 2008 global financial crisis.

The nation that built its middle-class dreams on coding is being ruthlessly dragged back to reality by AI.

/ 02 /

Why Did India Fail to Stay at the Table?

With its old rice bowl shattered, India—boasting so many tech-savvy engineers—could theoretically pivot to seize new opportunities in the AI era. The reality, however, is that India hasn't even secured a seat at the "cake-cutting table."

U.S. asset manager Altimeter calculates that global AI net profits reached $637 billion in 2026. The U.S. captured 49%, South Korea 35%—together gobbling up 84% of global AI profits.

The remaining 16% was divided among Taiwan, mainland China, Japan, Europe, and others. India doesn't even appear on this lengthy profit-sharing list.

Many attribute India's AI missed opportunity to insufficient policy investment or computing power shortages, but these are merely surface symptoms. The real issue lies deeply embedded in India's decades-long industrial trajectory.

Looking back, Japan, South Korea, and China all underwent arduous industrial upgrades. Japan progressed from low-cost to high-quality automobiles, then to semiconductor materials. China moved from contract manufacturing to consumer electronics, then to internet products and AI.

You'll notice that East Asian nations consistently "built things" at each stage. What was India's path? IT services, IT services, and more IT services. India skipped industrialization entirely, "leapfrogging" directly into services.

This wasn't due to innate aversion to manufacturing but to self-imposed institutional shackles. After independence in 1947, India implemented a bizarre "license raj" system—any factory construction, expansion, or product line change required central approval. This system essentially protected vested interests, denying newcomers licenses.

By the time reforms arrived in 1991, East Asian nations had already carved up the low-end manufacturing pie. With foreign exchange desperately scarce in the 1990s, India was "forced" into global IT outsourcing.

Having missed the global manufacturing supply chain division, India naturally became a spectator in the AI infrastructure boom.

If hardware was unattainable, what about software models? The obstacles proved equally insurmountable, following the same logic that caused India to miss the internet era.

During the internet age, the U.S. produced Google, Amazon, and Meta; China produced Alibaba, Tencent, and ByteDance. India? With the world's largest programmer population, it only nurtured outsourcing firms.

The core reason is that India's domestic market couldn't serve as a product iteration testing ground. A successful software company requires a flywheel effect: "large domestic market → economies of scale → product iteration → globalization."

While India boasts 969 million internet users, their spending power remains abysmal. Current per capita GDP stands at just $2,800, with extreme wealth concentration—approximately 228 million people still live below the poverty line.

Most Indian users' limited purchasing power makes it difficult for Indian internet firms to replicate the U.S.-China growth model of scaling through domestic market size.

U.S. and Chinese tech giants typically validate their business models domestically before expanding overseas.

Take the U.S. example: SaaS giants usually secure large numbers of high-paying domestic clients before going global. American enterprises are willing to purchase software and serve as early adopters, helping startups repeatedly refine their products.

The opposite was true in India. Most enterprises had low digitalization levels, were price-sensitive toward software, had fragmented procurement processes, and preferred custom development and human services. Consequently, Indian SaaS firms often had to compete in U.S. markets before completing product validation domestically.

This explains a peculiar phenomenon: despite nearly 20,000 SaaS companies—one-fifth of the global total—India has produced few unicorns.

If even asset-light internet platforms struggle to thrive, how could they support capital-intensive large language models burning billions annually?

Thus, Indian tech giants never invested in AI.

The top five outsourcing giants—TCS, Infosys, etc.—accounted for over $500 billion in combined market cap, yet maintained pitifully low R&D spending.

In FY2008-09, TCS allocated just 0.2% of revenue to R&D; Wipro allocated 0.19%. Fifteen years later, these ratios barely budged. In FY2025, TCS spent 1% on R&D, while Infosys and Wipro allocated 0.5%. For comparison, Microsoft spends 12%, Google 14%, and Meta a whopping 25% on R&D.

Where did all that profit go? To shareholders.

From FY2020-25, the top five IT firms returned approximately ₹4.8 trillion to shareholders—87% of their combined net profits. Such dividend payout ratios exceeding 80% are rare among global tech companies.

This stems from two crushing pressures: business conglomerates and Wall Street.

Take TCS: its parent company, the Tata Group, operates steel, automotive, and other capital-intensive, cyclical businesses employing tens of thousands. TCS must maintain high dividend payouts to funnel U.S. dollars back to the parent company to cover losses.

Meanwhile, Wall Street prices these outsourcing firms as high-dividend "bond-like" assets. If they dare reinvest profits into uncertain large model R&D, capital would vote with its feet.

Outsourcing giants must return profits to shareholders; capital markets reward stable cash flows with higher valuations. In this self-reinforcing cycle, no one has incentive—or courage—to innovate.

Ultimately, India's AI "absence" represents the inevitable fate of its industrial trajectory.

/ 03 /

When Cheap Labor Becomes a Burden

So what does India's AI industry look like today?

As of H1 2026, India has just three generally recognized (recognized) AI unicorns.

Sarvam AI is the only firm truly building foundational models, completing a $1.5 billion Series B in June 2026—but generated a paltry $5.4 million in FY26 revenue.

Krutrim once vowed to rival OpenAI but dismantled its AI assistant within two years, paused chip R&D, drastically downsized its team, and pivoted to selling AI cloud services—with 90% of revenue coming from "internal transfers" with its parent company.

The third, Neysa Networks, simply rents out computing power.

These three unicorns, developed with national resources, collectively value at under $4 billion. In the global AI gold rush, this barely qualifies as "entering the game."

The International Monetary Fund's (IMF) latest forecast seals this transformation: it downgraded India's FY2026-27 GDP growth forecast to 5.8%—the largest downward revision since the pandemic. The core reason cited was permanent contraction in the national IT services export pillar.

This isn't just an Indian story but a cruel economic parable.

When an economy treats "cheap labor" as its core competitive advantage for too long, building a vast benefit distribution (interest distribution) system around it, it essentially bets against technological progress. The faster technology evolves, the more severe its backlash becomes.

Holding the world's largest pool of cheap labor becomes the heaviest burden in the AI era. The nation with the most programmers finds itself locked in the twilight of the old age.

Text/Yuanyuan