#322 Between Winning and LosingAI Adoption, A Long-awaited Reform, and the Economics of Tech Substitutions
Applications for the 43rd cohort of Takshashila’s Graduate Certificate in Public Policy are now open. If you enjoy reading this newsletter, doing the GCPP is the next logical step.Table of Contents
India Policy Watch #1: Loser?Insights on current policy issues in India—RSJBack in April, I had written about my first-hand experience of AI adoption across a cross-section of companies in India. I had two early conclusions. One, Indian IT services is in for a serious disruption over the next three years, where it will have to rework its business model, shrink its revenues and then look for possible future growth. All quite painful to execute. Two, India will be a consumer of AI services rather than a core value creator. Being a consumer would mean it will use these services, provide a service wrapper over AI engines to serve clients, and possibly extract automation and productivity gains across sectors using these tools. In all these cases, the economic value of this will be split between the Indian firms and the global AI providers, with a larger share accruing to the latter. The value from being a consumer of AI will lead to productivity growth, but it is difficult to see where the excess human capital available in India will then be productively deployed. This will be a serious challenge to manage. On the other end of the spectrum, India will struggle to take part in the upstream value chain of the AI disruption, like chip design, manufacturing, creating competing foundational models or using AI to create disruptive products that will find global demand. India today lacks the ecosystem, R&D investment, and patient capital to play in this space. Even a focused effort driven by the state with active participation from all players will take a decade to be relevant here. The world would have moved on by then. About seven months later, as I read through strategy notes about India from research houses and hear from global investors, this thesis of India being an ‘AI loser’ is actively at work. There’s no real Indian innovation story that’s emerged lately. In most emerging sectors, India isn’t in control of the underlying technology. The rapid scale that China has built in green tech is a threat to the traditional areas of manufacturing strength of India, like auto parts, ancillaries and chemicals. In any case, India was not exactly capitalising on the “China+1” strategy that emerged from COVID-19. To be sure, the Indian MSME players have benefited from the broader trend of China diversification on a tactical basis. However, at a macro level, in multiple sectors, other smaller and nimbler economies, such as Vietnam, were the major gainers. China itself has shaken off its self-doubt on investment-led growth and doubled down on its manufacturing dominance with a specific focus on higher-end manufacturing and new technologies. In any case, even without the trend of de-globalisation and tariffs that’s prevalent now, the world would not have allowed India to become ‘another China’ in being the manufacturing hub of the world. The lessons from letting China get away with the benefits of globalisation while not playing by the rules of the game (currency manipulation, non-tariff barriers, etc, etc) have all been learnt now. So, even without this tag of being a net AI loser, it would have been tough for India to jostle its way to being an export-oriented powerhouse. The lack of a strong strategic lever that allows India to have a geoeconomic heft, barring access to its markets, was quite apparent when Trump bullied it on the trade deal. That is a real problem to contend with for India. The initial response from India to address this problem, as I see it, has been somewhat puzzling. There is a view emerging that India should resist being a consumer of global tech. Maybe that way it can avoid the fate of falling behind here. So the thinking is that it should have its indigenous home-grown technology players substitute the global giants to service its domestic needs. The recent decision by several government departments, ministers, and the PMO to relocate their mail servers to a domestic suite, in the name of data sovereignty and enhanced security, is a case in point. More such ideas seem to be in the pipeline. Quite honestly, this is mere optics to show the illusion of action and a flex on a technology that belongs to the last century. Trying to do the same on cutting-edge technology and foundational platforms will come to a cropper. I can see homegrown tech companies championing the case for this in a move reminiscent of the Bombay Plan of 1944. This is seductive and can quickly become a policy doctrine. “Why can’t we build our own AI platform? We are the most intelligent people in the world, after all.” We are pretty susceptible to the rhetoric of this kind. The other day, I read a report about the FM speaking to students at the Delhi School of Economics with some worrying advice to them (from the Mint):
I’m not sure if economic models are different for different geographies. “India-centric models, read between data and talk for India” - all seem to be a code for something that I can’t quite comprehend (the closest parallel I can think of is Erdogan running his own parallel economics theory and model back in 2022). Whatever it might be, it is not economics for sure. Protectionism and intellectual dishonesty are the easiest economic tools to reach for when things don’t look great. I can see more of this coming, and I’m afraid the outcome won’t be any different from what the Bombay Plan eventually achieved. The other counter to the AI loser thesis is that for the foreseeable future, the business cycle and the macros will trump such AI considerations. I agree to this to some extent. AI differential will take time to play out (over 5 years, even in the best-case scenario). Till then, doing the basics right will still yield the right results. Managing the fiscal deficit, supporting credit growth, deregulation, price stability, and pragmatism on trade relations are all levers available to policymakers to keep nominal annual GDP growth in excess of 11 per cent for the next five years. There is a possibility, in the meantime, of some kind of AI bubble bursting, greater realism about the value AI can deliver, more clarity on pricing for India to work on taking advantage of AI, bilateral trade deals working well and thinking through what strategic lever India can have for future negotiations. These can all happen in the interim. In essence, if we don’t screw up the business cycle that we might have for the next few years given our current reality, we get a good window to craft a long-term innovation strategy that addresses our status of an AI loser without a strong geoeconomic or geopolitical card in hand. What that innovation strategy might be is beyond me. But it should be a primary priority of policymakers at this moment. Perhaps they should read more of Mokyr. India Policy Watch #2: Electricity Reforms - Making it HappenInsights on current policy issues in India—Pranay KotasthaneBy now, we all know that high electricity tariffs for industrial and commercial use cases are a significant hindrance to manufacturing in India. A prime reason for these high rates is the implicit subsidy for electricity used to operate irrigation pumps. This is an implicit subsidy because the electric supply to agriculture is not metered in most states. Theft of power is often misclassified as agricultural use, and there is no way to measure the amount of theft. This waiver makes cross-subsidisation necessary, resulting in higher tariffs for other sectors that aren’t as politically powerful. Even so, state governments prevent the state electricity regulator from raising tariffs to a level truly reflective of the cost of provision because that might invite a pushback. Instead, state governments prefer that the distribution companies (DISCOMs) absorb the losses and remain in a comatose state of debt rollovers. Meanwhile, state governments don’t shy away from making electricity free for urban consumers as well. Why bother about fiscal responsibility when the debt can be kept off-budget, locked into a struggling DISCOM? This is the story of the electricity sector over the last three decades. Governments are well aware of this problem and have attempted several reforms, which primarily gave sops to states to take over the debt from the DISCOMS in return for some relaxation in the debt repayment terms. The union government often absorbed the restructuring burden, and in return, forced states to commit to reforms such as setting up State Electricity Regulatory Commissions (SERCs), unbundling their State Electricity Boards (SEBs) into separate generation, transmission, and distribution companies, and compulsory smart metering. These band-aid reforms haven’t stopped the DISCOMS from bleeding; the total losses stand at nearly ₹7 lakh crores. The problem was that governments developed cold feet at the idea of raising the rates or merely estimating the exact subsidies given to agriculturalists. The current NDA government has also tried reforms in this sector twice before, but couldn’t go further than kicking the can down the road. But the penny seems to have dropped. The Draft Electricity (Amendment) Bill 2025 has several foundational ideas, some of which are below:
Of course, this sounds too good to be true. There’s no reason why state governments that aren’t even ready to install meters for agricultural connections will implement any of these measures. To align their cognitive maps, the carrot being dangled is another debt restructuring plan. This would involve a new bailout package for states that agree to divest some equity from their DISCOMs or list them on a stock exchange. Let’s hope that this “carrot and stick” reform gets implemented. It has several features to like, and is not just another case of postponing the problem. However, making it successful requires at least four other features. One, sustained political will. These reforms will be politically painful. Tariffs might rise for some consumers, cheaters will have to pay, and the state’s budget will take a direct hit from paying subsidies in cash. Thus, the NDA government needs to push through these reforms in the states where it holds power first. Second, improved service quality. For consumers and political leaders to accept cost-reflective tariffs, they must see a benefit. The Bill enables the prescription of uniform “minimum standards of performance”. Enforcing these standards is crucial. If consumers are asked to pay more for the same poor service, the reforms will fail. Third, true regulatory independence must be realised. The Bill gives SERCs the power to act suo motu, and the threat of removal if they fail to do so. However, this needs to translate into functional independence. The process for investigating and removing a regulator for “gross negligence” must be robust, fair, and insulated from political influence, likely relying on the expanded Appellate Tribunal (APTEL) to make the final determination. Fourth, broader tax & structural reforms. Electricity is still outside the GST. This means power generators and distributors pay GST on their inputs (such as coal, equipment) but cannot claim an input tax credit, which inflates their costs. Bringing electricity under GST would break this cascading tax cycle, making the entire supply chain more efficient. For now, all we can say is that this is a crucial reform, much like the GST rate rationalisation, in the face of global economic turbulence. Electricity will be a key input for leading sectors—even more so than in the past—and hence getting it right is a pro-market reform that will go a long way in improving business prospects across sectors. Global Policy Watch: Technology ModulationGlobal issues relevant to India—Pranay KotasthaneThe narrative that China’s dominance in rare earth processing gives it a long-term geopolitical advantage continues to dominate the headlines. I have written on several occasions in these pages why I don’t buy that argument. The argument rests on the logic that substitutes for rare earths or rare earth-based components are around the corner, and China’s actions are making them economically viable. I am now willing to say that by December 2027, rare earths will no longer be a geopolitical lever for China to coerce other capable nation-states. Just last month, Flash Metals USA announced its intention to productise a new critical minerals recycling technology called Flash Joule Heating in early 2026. Developed by researchers at Rice University, this technique can extract REEs from rare earth magnet waste in seconds without the requirement for water or acids, which pollute the environment. This development led me to ponder: how do we estimate the pace of technological advancements? There is a well-known narrative suggesting that people tend to underestimate technology in the short term and overestimate it in the long term. But this narrative overlooks a crucial point: urgency dictates when a technology comes to fruition. Numerous examples support this. Take synthetic rubber, for instance. At one time, natural rubber was a canonical dual-use good—essential for military vehicles, aircraft, and regular tyres. As Ansgar Baums and Nicholas Butts write in their book Tech Cold War, ‘geopolitics of the 1930s was partly rubber politics’. But natural rubber for the whole world was exclusively sourced from Southeast Asia. Once Japan occupied that region, the West’s rubber supply was cut off. Mayhem! Japan had the trump card. But the West didn’t counter this leverage by growing its own rubber. The technology to produce synthetic rubber was invented in 1908 in Germany, but it failed to gain traction because natural rubber was the cheaper and more readily available alternative. It was the geopolitical crisis of the Second World War that finally prompted companies to adopt synthetic rubber quickly and on a mass scale. Another example is the COVID-19 pandemic. Initially, it was believed that the design, approval, and distribution of vaccination would take many years. Yet, the first vaccine was publicly available in a few countries within a year of the virus being identified. These two examples illustrate that with sufficient national motivation and focus by capable economies, technological advancements get expedited. Things that seemed irreplaceable get replaced by other substitutes faster than one would expect. I think China’s weaponisation of everything creates similar conditions to those seen with natural rubber and COVID-19. It is clear that China will leverage its manufacturing dependencies for geopolitical advantage over and over again. Diversifying away from Chinese supplies is not only a strategic imperative but also a business imperative. The choice is clear: either invest in diversification or wait to be hit by the next round of sanctions and denials of basic materials and non-strategic equipment. In either case, costs will increase. But in the first case, at least firms will stay in control of their business choices. In the second case, they are at the mercy of China’s geopolitical situation. Where governments failed to convince their companies to make this shift away from China, China is succeeding. HomeWorkReading and listening recommendations on public policy matters
|
