Leading Edge
A Reddit post from a software architect in Bangalore made news: after a four-day sprint where Claude overhauled a multi-app login system, work that normally took two months, he was ordered to lay off 45 people. That was 90% of the tech team. The three remaining architects would absorb the entire workload.
Whether the math holds perfectly matters less than what the thread reveals about a crumbling foundation. India’s rise into the global middle class ran substantially through one channel: performing tech labor more cheaply and competently than wealthier nations could do it at home. The entire mythology of the Indian engineering college being the the gateway to work for the likes of TCS, Infosys, and Wipro, depended on the cost gap between Bangalore and Boston. That wage arbitrage was the moat. India didn't need to be the absolute best, just reliably good and signficantly cheaper.
But AI changes the game by arbitraging labor itself, not just its cost. As one commenter noted, a 3 times increase in AI productivity can either mean doing three times more work, or firing two-thirds of the staff. The architect's company chose the latter, and they aren't alone. Other commenters chimed in with identical stories: a mutual fund startup founder who axed his dev team to pivot to an "AI-native" model, and another firm keeping teams at half-strength by forcing survivors to bridge the gap with AI tools. None of this requires AI to be flawless; it just has to be good enough and cheaper that the headcount it replaces.
Arbitrage historically assumed that hiring in rich countries would always remain the pricier alternative. AI breaks this formula entirely. A model requires no salary and no visa, competing with both sides of the old equation simultaneously. What replaces India's economic moat remains an open question. The thread offered no answers, only gallows humor, advice to upskill, and a grim prediction from a 20-year veteran: "order-taking" developers who merely implement assigned tasks are the first to go. Whether that is a isolated corporate consolidation or the leading edge of a global collapse is something no one in the comments, least of all the architect who just left the meeting, could say with any confidence.
Whether the math holds perfectly matters less than what the thread reveals about a crumbling foundation. India’s rise into the global middle class ran substantially through one channel: performing tech labor more cheaply and competently than wealthier nations could do it at home. The entire mythology of the Indian engineering college being the the gateway to work for the likes of TCS, Infosys, and Wipro, depended on the cost gap between Bangalore and Boston. That wage arbitrage was the moat. India didn't need to be the absolute best, just reliably good and signficantly cheaper.
But AI changes the game by arbitraging labor itself, not just its cost. As one commenter noted, a 3 times increase in AI productivity can either mean doing three times more work, or firing two-thirds of the staff. The architect's company chose the latter, and they aren't alone. Other commenters chimed in with identical stories: a mutual fund startup founder who axed his dev team to pivot to an "AI-native" model, and another firm keeping teams at half-strength by forcing survivors to bridge the gap with AI tools. None of this requires AI to be flawless; it just has to be good enough and cheaper that the headcount it replaces.
Arbitrage historically assumed that hiring in rich countries would always remain the pricier alternative. AI breaks this formula entirely. A model requires no salary and no visa, competing with both sides of the old equation simultaneously. What replaces India's economic moat remains an open question. The thread offered no answers, only gallows humor, advice to upskill, and a grim prediction from a 20-year veteran: "order-taking" developers who merely implement assigned tasks are the first to go. Whether that is a isolated corporate consolidation or the leading edge of a global collapse is something no one in the comments, least of all the architect who just left the meeting, could say with any confidence.