ET NEWS

Budget 2026 is a business-as-usual exercise that increases some outlays, announces some new schemes, corrects some duty perversions, contains the fiscal deficit, and undertakes to devolve resources recommended by the Finance Commission to states. It gets a thumbs up by normal standards.But these are not normal times. We are experiencing what Mark Carney at Davos called a rupture in the global order. Trump has demolished old certainties about geopolitical alliances and enmities. No erstwhile US ally - European, West Asian or Asian - can count on Marines rushing to their rescue in the event of hostilities that, just a year ago, would have triggered intervention by the world's most powerful military. Every nation is scrambling to enhance its capacity to defend itself.While never an ally of the US, India had grown accustomed to considering itself a partner, not just formally in a grouping like Quad but in a comprehensive campaign, in general, to pre-empt any military adventurism by China in the Indo-Pacific. India's rise as an economic and military power is widely understood as a necessary condition to keep China's rise peaceful. Trump has ploughed such assumptions deep underground.Russia has been a reliable source of support. But Russia is beholden to China, for supplying critical inputs to its war against Ukraine. China has consistently dragged its feet over settling the border dispute with India, builds up and arms Pakistan as a convenient force multiplier, and engages with India's other neighbours in South Asia to neutralise them, if not turn them against India.India has to build up its strategic capacity, double-quick: buy and build arms, rid its communications and weapons systems of dependence on Chinese electronics, create a satellite constellation in low-Earth orbit to monitor developments on the ground, spot and lock in on targets, and supply vital battlefield communications, including to drones used by individual soldiers or patrol units.Simultaneously, there is strategic and economic challenge from AI. It's essential for commercial entities to develop AI applications that ride on foundation models from Western tech firms, and even open-source models available from China. But AI is increasingly used for military purposes. For that, India cannot rely on any model that can be withheld or corrupted by an external agent. India must develop its own.AI is a product of the computing capacity derived    from armies of high-speed, parallel processing chips housed in data centres, and programming. India has neither. But it has the potential to create both.India's semiconductor missions are ridiculous. The only reason to build chips in India, rather than import them, is to shield ourselves from weaponised technology access of the kind the US has used against China. That purpose is not served by handing over billions of dollars to foreign chip companies to set up low-end fabs in India. That money is better spent on developing our own chip-making ecosystem.A thought-through semiconductor mission would start with identifying every single element of the chip-making ecosystem, assemblies in each one of them, sub-assemblies and components. Fund five startups to develop every single item in this array, ranging from extreme ultraviolet lithography machines, lenses used to focus the laser to etch grooves on silicon wafers, to the kit to deposit vaporised metal in grooves and clean up afterwards.Not all of them will deliver. Some funds would be swallowed by the well-connected. But in this process of reinventing the wheel, India would create indigenous capability to keep innovating beyond the existing frontiers in this field.Can India do this? China has. Foreign companies have set up high-end R&D centres in their GCCs in India, employing Indian talent. What prevents Indian enterprise from doing this? Only the missing vision, courage and funding. Instead of luring foreign chip companies to set up shop in India, let GoI lure Indian tech talent working at the cutting edge in many parts of the world, to lead startups that would deliver Indian AI.All this calls for large amounts of money, which cannot be found by business-as-usual budgets. Given the strategic imperative, extraordinary measures have to be taken. GoI must draw back from its populist trespass into state subjects, and mindless subsidies, such as that for urea, which depresses nutrient use efficiency and inhibits the spread of complex fertilisers.It must put an end to the ethanol-blending farce, which only serves to reduce the crop under oilseeds and increase prices of maize, chicken feed, eggs and chicken. More than half of ethanol is derived from grain, including rice, all grown with subsidised, energy-intensive water and fertiliser.When Lal Bahadur Shastri called on people to hand over their gold to finance the war thrust upon India, Indians responded. India today faces a similar existential crisis. Indians will respond with sacrifice - provided it is shared by all sections, and funds are utilised wisely.Reorienting the budget is politics, not economics. It calls for political courage, honesty of purpose, and articulate engagement with the public. Is that too much to expect?

The government has expanded the criteria for recognising entities as startups by doubling the turnover threshold to Rs 200 crore, according to a notification.A new recognition of 'Deep Tech Startup' has also been introduced for entities working on cutting-edge and breakthrough technologies, the notification of the Department for Promotion of Industry and Internal Trade (DPIIT) said.The age and turnover limit criteria for such Deep Tech Startups have also been significantly expanded.According to the notification, the age limit has been extended from 10 years to 20 years from the date of incorporation or registration, and the turnover limit has been enhanced to Rs 300 crore."This step addresses the unique requirements of deep tech entities operating in areas with long gestation periods, high R&D intensity, and capital-intensive development cycles," the DPIIT said.Further to support innovation-driven growth at the grassroots in agriculture, allied sectors, rural industries, and community-based enterprises, the government has extended startup recognition eligibility to cooperative enterprises.Accordingly, certain categories of cooperatives are now eligible for recognition as startups, subject to conditions.It includes multi-state cooperative societies registered under the Multi-State Cooperative Societies Act, 2002, and cooperative societies registered under State and Union Territory Cooperative Acts, it added."Keeping in view the evolving startup ecosystem and the need to support startups with targeted benefits at various stages of their business lifecycle, the turnover limit for recognition as a startup has been increased from Rs 100 crore to Rs 200 crore," it added.Over the last decade, India's startup ecosystem has shifted toward longer innovation cycles, higher capital intensity, and delayed commercialisation, particularly in deep technology, manufacturing, and R&D-driven sectors.Several innovation-led enterprises currently outgrow existing age or turnover limits while still in critical development or validation stages, resulting in premature loss of recognition and access to policy support.The decision follows extensive consultations held with various stakeholders in the startup ecosystem, as well as different ministries and departments.The updated criteria are expected to expand access to startup benefits for research and innovation-driven enterprises; support deep tech ventures requiring extended development timelines; enable cooperatives to lead innovation in agriculture, rural development and allied sectors, it said.It added that as Startup India enters its second decade, these reforms are intended to provide a more predictable, inclusive and future-ready policy environment for founders and to attract long-term patient capital into high-technology and R&D-intensive sectors.So far, about two lakh entities have been recognised as startups.Recognised startups are eligible for a number of incentives, such as income tax benefits under the Startup India initiative by the department.

Shares of InterGlobe Aviation ended nearly 1 per cent lower on Thursday after the Competition Commission ordered a detailed probe against IndiGo for unfair business practices.The stock dropped 3.65 per cent to Rs 4,782.45 during the day on the BSE. It later trimmed most of the early losses and ended at Rs 4,933.95, down 0.60 per cent.At the NSE, shares of the company ended at Rs 4,932.20, registering a drop of 0.57 per cent. The stock had declined 3.63 per cent to Rs 4,780.30 apiece in intra-day trade.The Competition Commission on Wednesday ordered a detailed probe against IndiGo for unfair business practices, nearly two months after the country's largest airline cancelled thousands of flights due to operational issues, causing hardships to passengers.After taking into consideration data related to airlines and those provided by the aviation regulator DGCA, the Competition Commission of India (CCI) has prima facie concluded that IndiGo has abused its dominant position.In a 16-page order, CCI said that by cancelling thousands of flights, which constituted a significant portion of the scheduled capacity, IndiGo effectively withheld its services from the market, creating an artificial scarcity, limiting consumer access to air travel during peak demand."Such conduct by a dominant enterprise may be viewed as restricting the provision of services under Section 4 (2) (b)(i) of the Act," the regulator said.Section 4 of the Competition Act pertains to abuse of dominant position.Noting that prima facie the airline's conduct seems to be causing an appreciable adverse effect on competition in India, CCI ordered a detailed investigation by its Director General (DG).

Rail Vikas Nigam Ltd (RVNL) reported a steady performance in the December quarter, with profit rising modestly, even as higher expenses limited margin expansion. The state-owned rail infrastructure company posted a net profit of Rs 324 crore in Q3FY26, up about 4% from Rs 312 crore in the same quarter last year. The improvement in profitability came despite marginal growth in topline and continued cost pressures.Revenue from operations increased 3% YoY to Rs 4,684 crore, compared with Rs 4,567 crore in Q3FY25. Including other income of Rs 252 crore, total income for the quarter stood at Rs 4,936 crore, higher than Rs 4,836 crore a year ago.Expenses, however, rose at a faster pace than revenue. Total expenses climbed to Rs 4,577 crore during the quarter, from Rs 4,480 crore in the year-ago period. Operating expenses increased to Rs 4,354 crore from Rs 4,219 crore, reflecting higher project execution costs.As a result, profit before tax came in at Rs 415 crore, broadly flat compared with Rs 413 crore in the corresponding quarter last year.On a nine-month basis, RVNL reported net profit of Rs 689 crore, lower than Rs 822 crore in the same period last year, while revenue from operations rose to Rs 13,716 crore from Rs 13,496 crore.Alongside the results, the board announced an interim dividend of Rs 1 per share, representing 10% of the paid-up equity share capital, for the financial year 2025–26. The company has fixed February 11 as the record date to determine shareholder eligibility, and the dividend will be paid on or before March 6.

Detailed, extensive, and honest – that is how the 2025-26 Economic Survey described the Indian AI ecosystem, which has yet to make a global or domestic mark. In this context, the survey also acknowledged the global dominance of foreign AI systems. Their dominance, coupled with their rising expansion costs, high resource use, entry barriers, and environmental impacts is now forcing a realignment of India’s AI ambitions. Rather than chasing a global frontier AI model the survey recommends focusing on developing multiple sector-specific, customized AI models. While a late-mover advantage has helped India recognize the “unsustainability” of frontier AI model growth, this delay now limits domestic AI development.According to Stanford’s AI Index Report, by 2024 the USA had the highest number of sovereign AI models at 40, followed by China with 15. This global dominance has translated into a deep penetration of foreign AI systems in India. Indians now account for the largest share of global mobile users of ChatGPT at 13.5%, as reported by Mary Meeker in 2025. At this scale, leading AI firms are able to exert substantial market power, placing intense pressure on key AI resources and raising barriers to entry for new players. This expansion has been accompanied by sharply rising scaling and environmental costs. Industry leaders, including the IBM CEO, have highlighted that escalating costs could challenge the feasibility of large-scale data centre expansion. The survey notes that some AI firms could spend almost half a trillion dollars by 2030 solely on compute infrastructure. On the environment side, a typical data centre consumes 3-5 million gallons of water per day, while a 2025 study by Harbin University and the University of Pittsburgh estimates that data centres could account for almost 10% of electricity demand growth by 2030.Given these realities, the survey suggests that India could benefit from avoiding the “costly path dependencies” of frontier AI models and instead focus on domestic, sector specific development. While the survey refers to this shift as a benefit of hindsight gained by being a late mover, this timing has allowed foreign AI firms to establish a strong presence in India, making the growth of domestic models more challenging. The easy availability of foreign LLMs has led to widespread adoption at the firm level. According to a survey by the Competition Commission of India, 67% of surveyed firms were already building applications using foreign-sourced foundational models. As the market moves towards a situation resembling a natural monopoly-like situation, large AI firms operating at scale can achieve relatively lower average costs. Given the substantial investments required to scale AI in India, a single well-resourced firm may be able to serve the market more efficiently than multiple smaller firms.Domestic development is further constrained by additional structural challenges, including brain drain, low R&D expenditure (0.6% of GDP in 2024, compared to 2.68% in China), uncertain access to GPUs, and semiconductor export restrictions imposed by foreign hubs. India also faces high fixed costs in expanding data-centre capacity. The country currently hosts only 274 data centres, according to Data Centre Map, compared with 364 in China and 3,959 in the United States.In this context, multi-sector AI development projects remain a challenging objective. However, this should not be interpreted as inevitable or as a reason for policy inaction. Rather, there is a pressing need to build on existing strengths while maintaining the development of domestic capacity as a long-term goal. This entails maximizing the benefits of existing AI systems by promoting adoption at the small-business level, encouraging wrapper-based and applied AI innovation, generating public sector demand, and strategically addressing constraints that may impede these efforts.A 2024 Salesforce report found that 78% of Indian small businesses were either using or experimenting with AI, but 41% of respondents expressed concerns about being left behind, and 60% found it challenging to keep pace with rapidly evolving technology. Some also felt they lacked the time to master all the technologies their company employs. These clearly create opportunity areas for policymakers to step in and assist with awareness, low-cost adoption, and employee skilling.Similarly, Indian startups are experiencing both opportunities and challenges, as many operate as “thin AI wrappers,” renting intelligence from foreign LLMs and embedding it within applications. When platforms introduce new features or adjust pricing, these startups face significant pressure. A survey by the Competition Commission of India found that only 23% of businesses have invested in proprietary training data or fine-tuned models. These challenges are further intensified by risk-averse investors who are reluctant to support experimental pilots. Consequently, there is a clear need to provide startups with safer investment options, greater freedom to experiment, and pathways to generate real value through AI wrappers. This can be facilitated through the development of data centres that support data training, and by leveraging India’s large pool of developer talent, domain expertise, and abundant domestic data. Additionally, the public sector can serve as a critical environment for the adoption of AI use cases in a more experiment-friendly setting.Beyond these measures, policymakers must also draw lessons from unrestricted foreign entry without sufficient domestic capacity. The erosion of AI sovereignty offers a cautionary parallel. Complete isolation is clearly not feasible, but regulation remains an option. Policymakers will have to walk a thin line between globalisation and domestic interests, which means that the expanding user base of foreign AI systems in India, along with schemes that encourage such trends, must be carefully re-evaluated.It must be noted that the solutions outlined here are not exhaustive since the rapidly evolving global AI ecosystem demands innovation and sustained effort. Additionally, to fully explore pathways for maximizing capacity and accelerating progress toward AI sovereignty, India requires active engagement from policymakers, industry leaders, startups, and the research community. Collaboration across these stakeholders will be essential to translate ambition into outcomes and ensure that domestic capabilities grow in both scale and impact.(Amit Kapoor is chair and Mohammad Saad is researcher with Institute for Competitiveness.)