Last week, Gujarat CM Bhupendra Patel met 24 startups at his Gandhinagar residence and announced a Sovereign AI Park through an MoU with Sarvam AI. It was more than a policy headline. It was a signal that the race to build startup ecosystems is entering a new phase. One where artificial intelligence is no longer just a sector to incubate, but the very engine doing the incubating.
Traditional startup incubation has always been resource-intensive — mentorship networks, physical infrastructure, investor matchmaking, regulatory handholding. These are activities that historically required large bureaucracies, significant capital, and years to scale. AI compresses all of this. It democratises access to expertise, automates compliance and due diligence, and — crucially — enables governments and incubators to identify the most promising ventures far earlier in their lifecycle.
India’s States Are Competing on AI Infrastructure
India’s federal structure is producing a healthy laboratory of experiments. Telangana launched a ₹1,000 crore startup fund-of-funds alongside India’s first Google for Startups Hub, pairing capital with the credibility of global tech ecosystems. Karnataka, home to Bengaluru’s deep talent pool, is pursuing a Startup Policy 2025–2030 with explicit targets to seed 25,000 new ventures — many beyond the capital, in Tier-2 cities like Mysuru and Hubballi. Tamil Nadu recorded a remarkable 36% CAGR in startup registrations over five years, with over half led by women founders. Maharashtra, meanwhile, has quietly become the nation’s data centre capital, laying the physical substrate upon which AI-native businesses depend.
What Gujarat is attempting — with its AI Innovation Challenge, structured support up to ₹1 crore per startup, and access to shared compute infrastructure — is a vertical integration of the entire incubation stack. Rather than leaving founders to piece together compute, mentorship, and market access independently, the state is assembling these as a sovereign service. India’s national AI white paper reinforced this direction, calling for a tenfold expansion in data centre capacity by 2030 and a centralised pool of GPUs available to researchers and startups at subsidised rates.
The Global Benchmark: From Gulf Ambition to Baltic Precision
Internationally, the contrasts are instructive. The UAE and Saudi Arabia are deploying capital at a scale that dwarfs most nations — Riyadh’s Project Transcendence commits $100 billion to AI infrastructure, while the UAE’s Stargate partnership with OpenAI, Nvidia and Microsoft will build a 1-gigawatt AI data centre campus. These are not startup policies; they are civilisational bets on compute as a form of sovereign power. For larger economies with abundant capital, this model works. But it is not the model most of the world can replicate.
Estonia offers the more universally applicable lesson. A country of 1.3 million people that has produced Skype, Wise, Bolt and Pipedrive, Estonia launched its Eesti.ai programme in January 2026 under Prime Minister Kristen Michal — with a target of growing the economy by 25% in five years. The programme’s design is telling: it is structured to operate like a startup itself, not a government ministry. It focuses on a small number of high-impact AI deployments across healthcare, education and governance, rather than scattering resources across hundreds of pilots. Its AI Leap initiative puts AI tools directly in the hands of teachers and high school students. And its AI Gov Stack — open-source AI components for government services — is explicitly designed for export to other nations.
The Real Opportunity: AI as Incubation Infrastructure
The convergence of these global examples points to a clear thesis: the most efficient startup ecosystems of the next decade will not be those with the most incubators, but those with the smartest AI-enabled infrastructure beneath them. AI can evaluate thousands of startup applications with consistency that no committee can match. It can monitor portfolio companies for early distress signals. It can connect a founder in Hooghly (West Bengal) with a relevant investor in Singapore or a potential customer in Amsterdam — at zero marginal cost. Perhaps most powerfully, it can translate the regulatory, legal and financial complexity that crushes early-stage founders into navigable, personalised guidance.
The question for policymakers is no longer whether to use AI in startup incubation — it is whether they build this infrastructure in-house or cede it to platforms they do not control. Gujarat’s insistence on a Sovereign AI Park, Estonia’s open-source Gov Stack, India’s national compute pool — all reflect the same underlying intuition: that AI infrastructure is too strategic to outsource. The states and nations that treat AI not merely as an industry vertical but as a foundational layer of their innovation economy will be the ones producing the next generation of globally competitive startups. The garage hasn’t changed. What’s changed is everything around it.
