New Delhi: Amid the rapidly expanding universe of business-to-business (B2B) artificial intelligence (AI) startups, enterprises are now increasingly willing to pay for projects rather than seek free pilots—a sign of the market's growing confidence and maturity. However, most startups are yet to secure the large, multi-year contracts needed to scale, setting the stage for consolidation in the sector.
The commercial shift coincides with faster enterprise adoption of AI. About 40% of Indian respondents reported significant or full AI usage, compared with a global average of about 28%, according to Deloitte's March report, ‘State of AI in the Enterprise 2026: India Insights’.
Experts say while the era of free pilots is “largely over”, most contracts remain limited to departmental deployments, scoped implementations or paid pilots, rather than multi-year, enterprise-wide contracts with predictable renewals.
“More than half of new engagements between enterprises and AI startups still begin as proof-of-concepts. However, the era of free pilots is largely over. Enterprises are increasingly willing to pay for pilots, provided there are clearly defined success metrics and a credible path to production deployment," said Sanjay Dawar, partner and leader, One Consulting, PwC India.
“Many Indian companies are still hesitant to adopt AI, viewing it as a passing hype. Multinational companies, on the other hand, understand the technology and see its value, but their local executives often lack the authority to onboard startups," Jayanth N. Kolla, founder & partner at Convergence Catalyst, an applied AI venture studio. "Decisions typically require multiple approvals, including from global teams, causing many deals to fall through. As a result, Indian AI startups are increasingly offering free pilots to win enterprise logos, build credibility, and prove the value of their products.”
The major challenge for startups is, thus, not about generating enterprise interest but to convert successful pilots into production deployments, experts said.
“The bigger challenge is not getting enterprises interested in AI; it is moving from an interesting pilot to a production deployment before the startup exhausts its time, resources, and compute costs," said Ankit Bose, head of AI at National Association of Software and Service Companies (Nasscom). "Long procurement cycles, extensive customization and unclear ownership continue to slow this transition.”
Consolidation and shutdowns
“The demand opportunity is substantial, but consolidation is also inevitable,” said Dawar of PwC India. He said funding concentration among a handful of market leaders, and enterprise preference for a few trusted AI vendors are making it harder for startups to scale.
“As a result, many of today's application layer AI startups are unlikely to exist in their current form three years from now. This should be viewed as a natural stage of market maturation, rather than a sign of weakness in the ecosystem,” he added.
For context, the number of startups that shut down in this space have increased from three in 2020 to 18 in 2025, with most of them shutting down in the seed stage itself, per Tracxn data.
“Some of the shutdowns reflect the previous wave of AI companies: computer vision, deep learning, NLP (natural language processing) or automation startups built before the current foundation-model era," said Kompella. "Some of them are pivoting. Some are shutting down and restarting. Some are finding that their old technology stack or old go-to-market model no longer fits the market. That should not automatically be read as a weakness in the current AI wave. It is also creative destruction.”
Analysts say India's AI opportunity remains significant, particularly as banks, insurers, information technology (IT) services firms, global capability centres (GCCs), retailers, healthcare providers and government agencies expand AI adoption. But succeeding will require more than just building another chatbot or automation tool.
Startups with proprietary data, deep domain expertise, enterprise-ready products and repeatable deployment models are expected to emerge as long-term winners. Those relying on generic AI capabilities may struggle as foundation models continue to improve and competition intensifies, experts said.
"Relying only on Indian customers may not be enough. Startups must expand into global markets and focus on specialized use cases, rather than generic AI services," said Mrinal Rai, an independent industry analyst.
"The real issue is that enterprise decision-makers are unwilling to take risks. We built AI platforms ourselves and, despite having a 15-year-old business in India, we went to Europe and the US to sell them. The response there was phenomenal, and the pilots were paid. That's why many experienced founders are now building products in India but selling them overseas, a trend people are calling the ‘Skip India’' movement," said Kolla of Convergence Catalyst that has built multiple AI platforms.