The decimation of tech and software stock valuations underlines the risks first stated by Alan Turing. This week, the market and IT shares came face to face with their Minsky moment, the tipping point when fears becomes a reality. In less than two trading days, India’s IT shares lost nearly `3 lakh crore of market cap and marquee stocks like TCS and Infosys lost nearly a third of their worth, pushing them back closer to 2022 levels.
It isn’t just in India. Globally, software and services companies lost trillions of dollars in market value—ServiceNow is down 28 percent year to date, Salesforce 26 percent and Intuit over 34 percent. Tech-Software exchange traded fund slid 20 percent and the MSCI World Software & Services Index was down over 12 percent. Punters coined new phrases—SaaSpocalypse, SaaSmageddon—to define the collapse. Software has been pulled down from the traders’ pulpit and relegated to being just another utility. The bullishness of 2023, when analysts goaded investors to load up on software stocks, has been replaced by the bearish ‘AI discount’.
Theory oftentimes waits for attention and tangible evidence. As early as 2017, computer scientist Ashish Vaswani, along with seven colleagues, told the world, ‘Attention is all you need.’ The transformative concept of ‘attention mechanism’ led to the launch of ChatGPT. The threat to human interface in software services has been a constant presence. Yet, nobody quite bet against legacy giants—after all, as Hemingway says in The Old Man and the Sea, “It is silly not to hope.” Claude Code replaced hope with pessimism and dread, resulting in a one-way trail in valuations.
The success of Claude Code and Claude Cowork illustrates the power of the mathematical breakthrough of 2017. The retrenching of human interface represents a structural shift. The migration—from the earlier autocomplete era to the automated engineer era—is tectonic in every sense. Consider this: earlier, the machine was a co-participant in a process enabling faster coding. Now the machine runs the code, sees the error and self-corrects until it works. The migration from CPU to GPU-based coding accelerates the shift from per-hour billing to outcome fees.
The fall in stock prices is the headline this week, but that isn’t the big news. IT and IT enabled services are India’s largest exporters, bringing in nearly $300 billion. It was also the largest formal employer till recently and is the force multiplier for consumption across sectors. It is useful to remember that just the software segment employs over 50 lakh people. Now the risk is two-fold—a slowdown in hiring and loss of employment. The quantum change is already visible. Soon after the advent of generative AI, a survey showed 80 percent of CEOs were planning to deploy AI—not just in services, but in manufacturing too.
The ability of machines to complete tasks has wiped out thousands of entry-level jobs across sectors—from software companies and law firms to consultancies and even creative shops. Unsurprisingly, in the US the unemployment rate for young workers (22-27 years) is over 7.8 percent. In India, the top five IT companies could altogether add just 17 recruits to their combined rolls in the first nine months of 2025. It is true that the transition could trigger demand in other areas—for instance, as AI architects. But it will drive down volumes and lead to a collapse in entry-level coding roles.
The entry of machines is already upending business models across sectors. The secondary effects of the slowdown or stalling of graduate hirings could well trigger a white-collar recession in consumption. For the record, since the launch of ChatGPT in 2022, the world has added over 1.1 billion AI users, over a fifth of them on ChatGPT.
The transition isn’t sui generis or driven only by private sector innovation. The push towards automation is also a geopolitical objective for the US. Post 9/11, America redesigned policies to enable energy independence. The Energy Policy Act of 2005 under George W Bush allowed tapping of shale oil and gas fields. The Energy Independence and Security Act of 2007 created a corpus of $25 billion for research. The 2009 American Recovery and Reinvestment Act under Barack Obama furthered energy independence. In 2017, thanks partly to fracking, the US emerged as the top oil producer.
The focus on automation and autonomous processes is not an accident either. It is the route to shrinking dependence on ‘imported labour’ and migration. The DARPA Grand Challenge in 2003 led to the creation, in 2006, of the CISE-IIS Division of Information and Intelligent Systems for research in human-computer interaction. In 2016, the US National Semiconductor Technology Center unveiled a strategic plan for “investments in the next generation of AI” and “develop effective methods for human-AI collaboration” through a network of AI institutes. The 2017 Tax Cuts and Jobs Act provided room for companies to invest by providing 100 percent depreciation—while wages are taxed (payroll tax), machines deliver tax cuts (depreciation).
This week, the government of India is hosting the AI Impact Summit. The guest list is impressive, but what is the focus? Can the ministers forge partnerships for AI-enabled solutions—gathering data for preventive healthcare, curbing air pollution, addressing teaching deficiencies, fixing farm linkages including using geospatial capacity to inform farmers, moving from flyover fixation to easing last-mile mobility in cities? There is no disputing that the fears of AI’s impact are real. This calls for rethinking policies. While doing so, it would be useful to remember ‘attention’ matters.
Read all columns by Shankkar Aiyar
Shankkar Aiyar
Author of The Gated Republic, Aadhaar: A Biometric History of India’s 12 Digit Revolution, and Accidental India
(shankkar.aiyar@gmail.com)