India

The word covers too much. A kirana, a services exporter with a thousand employees, a campus company, and a venture-backed product firm are different animals. Only the last is what this vertical means by a startup, and even that firm usually sits on top of a much larger services and trading economy.

Family capital and promoter control remain ordinary. A startup that takes outside money is choosing a different governance habit, not joining a national type. Caste, language, and city networks still decide who gets the first customers. Pretending the market is a single meritocratic funnel is a research error.

The working vehicles are a private limited company and a limited liability partnership, both on the Ministry of Corporate Affairs register. A partnership firm can also qualify for Startup India recognition. Recognition is optional. Incorporation is the legal fact.

Founders often add a foreign holding company so that overseas investors can take a share under a law those investors already know. That choice creates a second company, a second tax file, and a foreign-exchange question about how money moves into the Indian operating company. It is a structure, not a trick, and it does not replace Indian company law for the entity that employs people.

Domestic venture funds that pool money from Indian and foreign investors generally sit inside SEBI’s AIF regulations. The regulations have been amended repeatedly. The text read for this page is the version SEBI marks as last amended in November 2025. Category labels and investment conditions are in the regulations. They are not repeated here.

Angel money, corporate venture, and family offices exist beside that regime and are not the same licence. GIFT City, under IFSCA, is where a fund can be managed inside India on an international-centre rulebook. It does not move the operating company’s customers, staff, or state licences to Gujarat. Foreign venture capital still has to enter under the foreign-exchange and FDI framework the Reserve Bank and the government maintain. That framework is a living document.

The deepest public technology is not a startup. UPI moves money between bank accounts through private applications. Aadhaar is an identity system. Account aggregators and public digital goods sit in the same family: state or quasi-state rails that private firms are allowed to build on. A company that wraps a rail is not the owner of the rail.

The large technology employer in India is still export services: implementation, support, and engineering for clients abroad. Product companies hire from that labour pool and compete with it for people. Hardware, semiconductors, and laboratory science are real and much thinner. A services exporter and a product company hire for different work.

English is a working language in the firms this page is about, which is not true of the whole labour market. Engineering colleges produce volume. Quality and willingness to join an unproven company vary sharply by city and by the outside option of a services salary.

Returnees from the United States and from other Asian offices matter in some product companies and are irrelevant in others. A hiring plan that assumes a national market for senior product managers will discover it is a few cities, and that those cities do not share a cost of living.

The Indian Institutes of Technology, the Indian Institute of Science in Bengaluru, the Indian Institutes of Management, and the IIITs are public institutions with different charters. They produce graduates and, in some departments, research. They do not have a duty to produce companies.

Most founders did not come from the most famous campus. Treating IIT Bombay or IISc as the ecosystem is a way of ignoring Hyderabad, Pune, Chennai, and the private engineering colleges that staff the services industry. Those colleges are uneven. Uneven is not the same as irrelevant.

Company law is central. Labour, land, and many licences are state subjects or are administered locally. A firm that is lawful in Karnataka is not thereby finished with Haryana or Uttar Pradesh. GST replaced a tangle of indirect taxes with a different tangle. It is national. It is not simple.

Sector regulators matter more than the companies act once the product touches money, health, telecoms, or insurance. The Reserve Bank, SEBI, IRDAI, and TRAI each run their own process. FDI policy is consolidated and then amended. A paragraph in a foreign brief that states ‘the cap’ without a date is already stale. This page refuses the cap.

The World Bank discontinued the Doing Business report in September 2021 after data irregularities. GRIP does not use those country ranks, for India or for anyone else. Cost here means rent, payroll, professional fees, and the time senior people spend on filings.

Those costs are city facts. A Bengaluru engineering salary and a Hyderabad engineering salary are not interchangeable, and neither is a Mumbai rent. Compliance cost is also a stage fact: a twenty-person company and a company with a foreign holding structure do not have the same advisor bill. No index on this page will rank the cities.

Digital identity and retail payments are ahead of urban transport and, in some cities, water and reliable local roads. A product that assumes UPI can reach a customer a logistics network cannot. Power is better than it was in the services campuses and still an operating assumption outside them.

Airports connect the main cities to each other and to the Gulf and to Singapore more easily than the cities connect to their own peripheries at a weekday hour. For a software firm that is an inconvenience. For a company that moves goods, it is the business.

The domestic market is large, price-sensitive, and multilingual. A product that works for an English-speaking professional in one metro can fail in the next state for reasons that have nothing to do with the software. Public procurement is a separate market, with its own payment habits.

Export of services remains the path that brings foreign currency without winning a consumer. That path trains managers and also pulls the best operators toward client work. Selling to Indian enterprises, to the state, and to households are three go-to-market problems. Scale-up stories that collapse them are not research.

The distinctive innovation of the last decade is institutional: public rails for identity and payments, which let private companies start further along. Process innovation in logistics, credit assessment, and low-cost delivery is more common than a new scientific platform.

Laboratory science is concentrated and does not automatically become a company. Where it does, the constraint is usually a long regulatory path. Calling every consumer app ‘deep tech’ empties the term. DPIIT’s own recognition rules now distinguish a deep-technology window. The distinction is administrative. It is still a useful warning against the empty term.

A city product becomes a national company only if distribution, language, state rules, and working capital allow it. Payments infrastructure removes one excuse. It does not remove warehouses, collection, or a state licence.

Public markets exist. BSE and NSE are the securities infrastructure, under SEBI’s listing rules. Most startups will not list, and a listing is not a research conclusion. The alternative exit, a sale to a strategic buyer or to a later fund, depends on foreign-exchange rules if the buyer is abroad. Scale, in India, is usually a domestic operating problem before it is a capital-markets event.