Startup India at 10: With little to show, how failure was defined as success

Startup India was launched in January 2016 with an ambitious promise: India would stop waiting for technological futures to arrive from elsewhere and begin producing them at scale. The aversion to risk-taking and failure would be reimagined as part of innovation.

Ten years later, the headlines are impressive: over two lakh recognised startups, more than 125 unicorns – as startups valued at over $1 billion are known – and India’s position as the world’s third-largest startup ecosystem, says a government press release.

Startup India has also reshaped the collective imagination of techno-futures: the Indian version of the US television show, Shark Tank, entered its fifth season while tech entrepreneurs are being described as “wealth creators”.

But it is harder to explain what has not emerged.

India has not produced a transformative technology that reshapes global markets on its own terms. Most unicorn startups remain unprofitable. Most patents, filed to impress investors, were rejected in the first round.

Much-hyped Digital Public Infrastructures, such as the UPI payment system, continue to rely on state subsidies, with limited global uptake. Despite the scale of talent, data and users, India remains peripheral to frontier AI development. The standard explanation is that India is still “catching up”.

But if Startup India’s goal was to catch up with venture capitalism’s model of futurity as exemplified by Silicon Valley tech firms and investors, with imagined futures shaping actions and decisions in the present, it has succeeded in mastering one of its defining features: failure.

Globally, over 90% of startups fail. Startup India has not merely generated thousands of state-supported startups, many of which have failed. It has also institutionalised a relationship to futurity in which failure is endlessly displaced and managed without ever becoming instructive.

The first decade of Startup India can be read as four phases of the successful governance of failure.

Leaps and bounds

First, from 2016 to 2019, Startup India focused on rapid ecosystem-building: incubators or organisations that support early-stage firms, seed schemes offering financial support, recognition frameworks that motivate participants and state-wise rankings systems to monitor regional disparities.

Innovation became something that was visible to the government and trackable. For instance, the Startup India website tracks how many companies are recognised by the Department for Promotion of Industry and Internal Trade, and uses the Bharat Startup Knowledge Access Registry, or BHASKAR, for collaboration.

The wager was that once the machinery existed, innovation would follow. What went largely unquestioned was the imagination embedded in that machinery. Innovation was defined through a borrowed horizon in which venture capital judged value, rapid scaling was a sign of merit, and platforms and exits defined success – where building digital and social platforms, however unprofitable, and the exit of investors like venture capitalists and angel investors, is the main criteria for success.

Tier-2 and -3 cities were folded into this vision not as sites of alternative experimentation but as future replicas of metropolitan ecosystems that account for 90% of investments.

When startups and incubators struggled outside Bengaluru or the Delhi-National Capital Region, explanations rarely pointed to institutional support such as weak research and development ecosystems. Instead, failure was attributed to culture, mindset or entrepreneurial immaturity. Failure disciplined those who were considered unprepared for the future, while the future itself remained unquestioned.

Second, the Covid-19 pandemic briefly threatened this narrative as startups stalled, demand collapsed and uncertainty spread. But that moment passed quickly. By 2021-’22, the pandemic had become an accelerant rather than a rupture, with India briefly overtaking China in notching up new unicorns. National Startup Day was instituted in 2022, and the prime minister hailed startups as the “backbone of New India”.

In the pandemic crisis, EdTech platforms such as Byju’s substituted for schools, the government’s health app AarogyaSetu hid the collapse of public health while betting apps normalised largescale speculation.

Governance constraints due to the pandemic blurred questions of sustainability, labour conditions and institutional capacity. Instead of building durable systems, the state drifted towards brokering rapid digital fixes through opaque public-private arrangements powered by India Stack – a set of technologies that verify identities digitally and enables online transactions.

This can be seen in the launch of a new government EdTech platform, Digital Infrastructure for Knowledge Sharing App, or DIKSHA, developed by tech entrepreneur Nandan Nilekani’s foundation. Similarly, Bharat Health Stack was made by “volunteers” at the tech lobbying group iSpirt – the Indian Software Products Round Table.

This acceleration was underwritten by zero-interest-rate global capital that flooded technology markets, allowing speculation to stand in for innovation. Hiring spiked, valuations swelled and expansion was framed as inevitability. Failure in this phase was drowned out by acceleration. Burn, churn and precarity were absorbed into crisis rhetoric, while deeper costs were treated as temporary frictions. The digital-first future appeared to arrive early, albeit as a blip.

Third, from 2023, as capital tightened globally, the ecosystem entered what many called a reality check. Funding slowed, layoffs followed and profitability returned as a moral demand. Yet this reckoning did not challenge the imagination of innovation over the previous decade. Instead, it narrowed it further.

Startups doubled down on what monetised fastest, such as quick commerce servicing a tiny urban elite by exploiting gig workers. Problems requiring long-term institutional commitment remained structurally outside the startup horizon. Predatory “decacorns”, such as Byju’s, and thousands of small startups, vanished.

This churn was labelled a necessary attrition. Startups failed, but the future held promise. Each shutdown cleared the ground for the next cohort, the next pitch, the next promise. By treating failure as a corrective signal rather than a structural outcome, much contemporary sobriety about startups reinforced borrowed ideas of technological futurity.

Finally, the current AI moment condenses this history into a single frame. Once again, the language is about races and catching up. Once again, the architectures are borrowed.

Most generative AI systems begin from foundation models trained on massive, predominantly English-language datasets. Local adaptation occurs through fine-tuning and translation layers that import assumptions about language, value and users. Linguistic plurality becomes an engineering constraint rather than a premise for reimagining human-machine relations from the ground up.

When systems underperform across Indian languages and contexts, the explanation returns to data scarcity or technical immaturity. The deeper issue, that derivative horizons were adopted from the beginning, remains untouched.

This oxymoronic endeavour to build “sovereign AI” using Nvidia’s chips and wrappers on foundational models epitomises Startup India’s success in governing failure. Mega-events such as the AI Impact summit inaugurated on February 16 exemplify this. Instead of opening new possibilities, this approach helps maintain publicity cycles of what journalist Ravish Kumar once called “eventocracy”.

No lessons from failure

Across the four phases, failure has been individualised, regionalised or temporally deferred. It has rarely been allowed to grow into institutional learning. This poverty of the imagination is not accidental. Dominant ideas of innovation, merit and risk in India are deeply shaped by caste, where only a tiny, upper-caste elite can afford entrepreneurial risk-taking.

For others, failure is seen as a lack of merit or readiness. Instead of dismantling this hierarchy, Startup India has repackaged it such that innovation became a performance of the elite’s capacity to copy Silicon Valley at scale, while much of the country appeared mainly as labour, data, or untapped markets.

In this system, failure sorts India into a consumption hierarchy coded by a Maslow-meets-Manusmriti fantasy – much like American psychologist Abraham Maslow’s linear pyramid of needs, for India’s racialised and caste-shaped startup ecosystem, the bottom of the pyramid that they call “India 3,” is a “sub-Saharan” population outside the formal digital economy. This closes the possibility of redesigning the techno-social order.

Seen as a short history, Startup India’s most consequential legacy is the normalisation of a techno-futurity in which failure is made governable: it can be absorbed without reckoning and recycled without reimagining.

The argument here is not that India has failed to catch up with the US or China. It is that letting the future be primarily defined by “catching up” guarantees a particular kind of failure, in which imported models enrich a few while the future repeatedly misses its democratic potentials.

If India simply reproduces exhausted horizons at scale, the world does not gain an alternative future. It gains a larger version of already existing failures.

Sandeep Mertia is a professor of Science, Technology, and Society at Stevens Institute of Technology in the USA. He is writing a book on Startup India.