#331 Fast Brain, Slow Brain

Decoding the Rupee Depreciation, Geopolitics of AI Chips, and What Kind of a Great Power Would India Be?

India Policy Watch: (Mis)Reading the Tea Leaves

Insights on current policy issues in India

—RSJ

The Indian Rupee crossed the 92 level on Thursday before the central bank intervened. This kind of heavy lifting by the RBI to shore up the rupee has become a regular feature since April of last year. Foreign capital flow has been net negative in this fiscal year, and export performance remains weak, given the multiple headwinds to global trade. It is easy to brand this balance of payment weakness as cyclical, but that would be wrong.

The Rupee has been the worst-performing major currency in the past year, falling almost twice as much as its emerging market peers. There is more to this, therefore, than just the cycle hurting everyone equally. Much is made about how this round of sharp Rupee depreciation is different from the past because of strong economic growth, good fundamentals and a healthy forex reserve. And the argument therefore goes that we shouldn’t worry much about the Rupee so long as we are in the goldilocks zone, or we are an oasis of calm (take your pick of the metaphors) among all major economies in the world. So, as the Economic Survey 2026 notes:

“The paradox of 2025 is that India’s strongest macroeconomic performance in decades has collided with a global system that no longer rewards macroeconomic success with currency stability, capital inflows, or strategic insulation.”

Is this true? Or is it a convenient case of blaming the “system” and calling a difficult problem a paradox to avoid finding real solutions?

Let’s look at capital inflow. There is no doubt that all the Trump shenanigans have meant a flight to safety to the dollar from the emerging markets. This outflow has been more marked from India as it has come under direct Trump fire on tariffs, purchase of Russian oil and a delayed Indo-US trade deal, which just doesn’t seem to find the finish line. The optimists believe that once the deal is struck, a lot of capital waiting on the sidelines will flow in. I’m sceptical about it.

Firstly, the FPIs who have been net sellers for the last 12 months aren’t returning in a hurry. The net FPI outflow figures of last year mask the significant investment that came into the Indian debt market after its inclusion in global bond indices and a change in weights. That one-off won’t persist going forward, and while I sense some FPI equity investment will return in FY27, it won’t be enough to make up for the relative reduction in debt investment inflow. As much as we love listening to the homilies from global CEOs and fund managers visiting India about the strong domestic market, consumption-led growth and economic resilience, it has translated to very little real follow-through action.

We might not like hearing it, but investors view India in a geopolitical tight spot, a zugzwang of sorts, with limited manoeuvrability. The FTAs and trade deals will help obviate some of this stress, but they will continue to be thwarted by the desire across nations to revive local manufacturing, especially in strategic sectors, and protect domestic markets. This instinct is high in India, with all kinds of definitions of swadeshi being touted to justify protectionism, which will only invite a tit-for-tat response.

The other factor, possibly short-term in nature, is the FPIs favouring “AI winners” in their fund management strategy, which will weigh on their India allocation. This might not sit well with many who think there’s no reason to consider India as an “AI loser” at such an early stage of its evolution. They may have a point, but that’s not how equity capital is flowing at the moment. Building upstream AI capabilities and investing in innovation is almost impossible at this moment for India. Those costs are prohibitive, and there is already an asset bubble building there with insane valuations. Interestingly, the Economic Survey concurs with this:

”Attempting to close this gap would involve prohibitive fiscal costs towards what is becoming an unsustainable approach to AI development.”

The problem is figuring out what the downstream leadership in being an “AI consumer” in various sectors actually means and how that translates to net value creation. This clarity will take time to emerge. It is possible that India might create firms, or its traditional IT services strength will translate to dominance in the “AI user” economy worldwide, but foreign capital isn’t going to be convinced about it anytime soon.

On FDI, things aren’t different. There is good momentum in areas where global firms are pursuing a ‘China+1’ strategy and diversifying their supply chain and manufacturing base. This is a classic case of India being the best alternative to China in terms on cost and talent at scale. The need for India to integrate with the global supply chain and make it easier for such firms to set up plants in India is understood.

But beyond this, there are challenges. There’s been a gradual erosion of confidence among global companies that serve the Indian domestic market to build further capacity in India. The increasing monopolistic nature of multiple sectors in India that are dominated by the same set of 2-3 players with local knowledge and ability to work policy in their favour has meant a certain reluctance for foreign capital to enter those sectors.

The continued expansion of the footprint of these conglomerates across multiple sectors has meant a gradual closing of the doors for MNCs to invest in expansion. These domestic conglomerates by themselves aren’t keen on entering global markets and increasing India’s export competitiveness. They are happier doing the easier job of muscling their way into other sectors within India and using their clout to buy out or outprice existing local players.

The MSME sector, which has done well in the past four years, helped by the formalisation of credit (Udyam and GST information have been game changers for credit offtake), has benefited from the diversification away from China by global companies. But there is a limit to this beyond which the expectation will be for these companies to scale big and build significant capacity.

The problem is that China hasn’t remained still, watching this diversification of the supply chain away from it since COVID-19. The results of the significant increase in capacity during this time are being seen now, with China ramping up its production and outpricing the nearest competitors in multiple industrial segments. The record $1 trillion plus record surplus that it has seen in the previous quarter is the clearest sign of this.

Now, which MSME manufacturer in India wants to take a long-term capex investment and risk to build a large factory and then get totally outpriced by China in future? It is easy to say that Indian corporates have been given tax cuts, PLIs and faster credit, so it is now their turn to repose faith and increase capex. But who wants to be saddled with debt on their balance sheet and a rampaging China dumping their goods all around the world?

There are other factors involved, too. With the US wrecking the global trade order it built over the past eight decades, there is a belated recognition of the merits of being self-sufficient among its so-called middle power allies. Defence budgets are being shored up, energy self-sufficiency is being explored, and manufacturing is being revived.

This turn to swadeshi worldwide has given its Indian proponents new wings. There was a hint in the past six months that India was unwinding some of the trade barriers it had put up in the past six years in the form of duties and QCOs on intermediates, which would eventually help with exports. But I suspect the global turn to atmanirbharta can easily be used to name every other sector ‘strategic’ and then raise trade barriers on imports to protect it. Eventually, that will become a tax on exports and reduce the competitiveness of domestic firms.

So, the business of defending the rupee won’t stop unless we address the twin and interlinked challenges of the flow of foreign capital and raising exports. These challenges, as I have pointed out above, are more structural than cyclical. The impact of supporting the Rupee isn’t only about depletion of forex reserves and a steeper import bill that brings along inflation. The real problem of defending the Rupee is how much of liquidity is sucked out of the system to support it.

This is what has happened in the past nine months. The RBI has actively injected liquidity into the system with multiple open market operations (OMO) buying G-secs from banks. But that liquidity has been sucked out by its defence of the Rupee during this time. The tight liquidity will bite the banking system. It hasn’t done so far this year because, for long stretches, the liquidity has been positive. But it eventually will, and what it will do is starve the system of credit. We have seen this through 2024-25, where persistent liquidity deficit killed credit offtake and that led to a slowdown in growth.

The system liquidity is quite tight already, and given the pressure on the Rupee, I foresee a vexed challenge on hand for RBI to keep its promise of durable surplus liquidity to support credit growth while managing to keep the Rupee under 92. The only real solution is for capital inflow to come back. That will take time. The India consumption story is often taken for granted. It isn’t. It runs on the wheels of credit, which is greased in turn by system liquidity. Take that away, and it will sputter.

There’s an old joke about statistical average - that of your head being on fire while your feet are on ice, but you shouldn’t complain since your average temperature is just right. As much as I’d like to believe India is in the Goldilocks zone of being at the just right temperature, there’s a possibility that we are misreading the average.

---

Global Policy Watch: Locking Your Geniuses in Datacentres

Global issues relevant to India

—Pranay Kotasthane

Daro Amodei, the CEO of Anthropic, posted an important essay earlier this week. Titled The Adolescence of Technology, the novella identifies four major risks posed by powerful AI systems and goes on to identify defences against them. It is an influential essay worth reading and thinking through.

But for this post, I want to focus on the geopolitical aspects. Amodei is in the camp that strongly bats for chip export controls against China, and for the US “arming other democracies” with AI but doing so carefully and within limits. This paragraph summarises his view on how to maintain the lead against China:

First, we should absolutely not be selling chips, chip-making tools, or datacenters to the CCP. Chips and chip-making tools are the single greatest bottleneck to powerful AI, and blocking them is a simple but extremely effective measure, perhaps the most important single action we can take. It makes no sense to sell the CCP the tools with which to build an AI totalitarian state and possibly conquer us militarily. A number of complicated arguments are made to justify such sales, such as the idea that “spreading our tech stack around the world” allows “America to win” in some general, unspecified economic battle. In my view, this is like selling nuclear weapons to North Korea and then bragging that the missile casings are made by Boeing and so the US is “winning.” China is several years behind the US in their ability to produce frontier chips in quantity, and the critical period for building the country of geniuses in a datacenter is very likely to be within those next several years.³⁵There is no reason to give a giant boost to their AI industry during this critical period.

Geopolitics or chips doesn’t seem to be the essay’s strong suit. First, GPUs are nothing like nuclear weapons. An Nvidia H200 GPU is the size of a mithaai dabba and can be easily transported, smuggled, or accessed remotely. Moreover, the vast majority of applications of the H200 GPU are in the private sector. Unlike nuclear weapons, they can be ordered by any company that has $ 30,000 to spare. And unlike nuclear weapons, which only have destructive uses, GPUs are important for companies to take advantage of the many benefits of AI that Amodei himself lists.

More fundamentally, such arguments suggest America’s technologists and analysts have much less confidence in the American technology ecosystem than I do. For if you think GPUs are the thing that is stopping China from getting ahead of you, you're already cooked. You’ve lost the game because, given their widespread commercial applications, a motivated adversary will always find ways to access these chips regardless of your sanctions.

Consider the case of the H200 chips. There are roughly 700,000 such chips in Nvidia’s inventory. It would roughly take 2000 of these to train a DeepSeek-v3 model in under two months. There’s just no way of preventing the PLA or CCP from accessing 2000 mithaai dabbas from other countries if that is the only thing preventing them from gaining access to the supposed game-changing technology of the century.

To some extent, I can understand the logic behind targeted restrictions on China's purchase of advanced semiconductor manufacturing equipment. These are huge machines used only by a handful of companies worldwide. They are easy to track and control. But export controls on chips, really?

I thought the US's strength was its excellent technology ecosystem, its ability to attract the world's best talent, and its power to mobilise capital in innovative ways. Instead of doubling down on these strengths, the squabbling over chips indicates weakness and insecurity, not confidence. China restricts rare earths; the US restricts chips. What's the difference between the two then?

Sure, blocking chips can buy some breathing time for American companies building AI models. That perhaps is the primary motivation. Fair enough. But if the stated goal is to prevent an adversary State from accessing frontier AI, chips are hardly the bottleneck. Just as you can make a less-efficient motor without Chinese rare-earth magnets, the PLA can make powerful military AI systems using slightly less-powerful chips.

The other unsaid part of Amodei’s argument is that it subtly advises the US government to control the flow of chips not just to China but also to other capable democracies. As he writes:

I am supportive of arming democracies with the tools needed to defeat autocracies in the age of AI—I simply don’t think there is any other way. But we cannot ignore the potential for abuse of these technologies by democratic governments themselves. Democracies normally have safeguards that prevent their military and intelligence apparatus from being turned inwards against their own population,³¹ but because AI tools require so few people to operate, there is potential for them to circumvent these safeguards and the norms that support them. It is also worth noting that some of these safeguards are already gradually eroding in some democracies. Thus, we should arm democracies with AI, but we should do so carefully and within limits: they are the immune system we need to fight autocracies, but like the immune system, there is some risk of them turning on us and becoming a threat themselves.

All good words. Who could disagree? Except when you think that the “we” here means the US administration, which is currently hell-bent on antagonising major democracies like the EU, India, and Canada. The argument for “arming” democracies is inherently political and will depend on US policy orientation, which is hardly predictable at the moment. Thus, a much better approach is to keep the State at bay and leave GPU buy-sell decisions to companies alone. AI—like other general purpose technologies—is not a finite, zero-sum game. Don’t make it one.

PS: I wonder how this essay’s arguments would change if and when Amazon (a major investor in Anthropic) allows the physical sale of its own Trainium chips. Would it be okay to lose out on the lucrative China market? I guess not.

---

Matsyanyaaya: What Kind of a Great Power Would India Be?

Big fish eating small fish = Foreign Policy in action

—Pranay Kotasthane

With Trump decimating American authority, the mask is off. Many past American misdeeds around the world are coming to light, and the US is fast losing its narrative superiority. In such a scenario, a question worth pondering is: hypothetically, would India behave any differently if it had the power of the US? Are all great powers destined to misbehave, or is the US an outlier? It is a hypothetical question, and India is a long way from becoming a great power, but there are no taxes yet, on letting our imagination run wild.

A crude way to answer this question is to look at India’s conduct in a domain in which it is already a superpower and see how it has deployed its influence. And there is one domain that gives us a preview of India’s dominance: cricket.

India’s supremacy in cricket is structural. The BCCI receives approximately 40 per cent of the ICC’s annual revenue. England, the next highest, gets just 7 per cent. India accounts for roughly 85 per cent of the ICC’s global television revenues. The BCCI is wealthier than all other cricket boards combined. If international cricket were the international order, India would already be the hegemon. So what has it done with this dominance?

On the positive side, India’s support for Afghanistan's cricket stands as a genuinely generous act of sporting diplomacy. When Afghanistan needed a home, the BCCI provided one—the Shaheed Vijay Singh Pathik Sports Complex in Greater Noida became their home ground in 2016, followed by venues in Dehradun and Lucknow. Former Indian cricketers coached the Afghan national team in various capacities. India contributed to the construction of a cricket stadium in Kandahar. Even the Taliban’s spokesman publicly thanked India for its “continuous help in capacity building.” Similarly, I am sure the BCCI has contributed to cricket in Nepal and in Bangladesh a couple of decades ago.

This is what India’s benevolent hegemony could look like: using structural power to elevate those who have none.

But here’s the thing: beyond Afghanistan, the benevolence largely disappears. Consider what India has not done with its unmatched power. There are 23 African nations that are members of the ICC through the Africa Cricket Association. Kenya once played a World Cup semi-final. But India's engagement with African cricket development has been negligible. There is no visible systematic effort by the BCCI to transform cricket in regions where the sport could genuinely grow. BCCI itself is obsessed with the traditional “big four” of cricket—Australia, England, South Africa—and what happens between them.

If anything, India’s conduct in the past year reveals something more troubling than mere neglect. It reveals the instinct to weaponise sporting dominance. At the Asia Cup in September, the Indian team refused to shake hands with Pakistani players across all three matches. The no-handshake policy has since been extended to India’s women’s and Under-19 teams.

The Bangladesh situation is even more instructive. In January 2026, the BCCI directed Kolkata Knight Riders to release Mustafizur Rahman, due to “developments all around”—a euphemism for deteriorating political relations. Bangladesh responded by banning IPL broadcasts and refusing to play T20 World Cup matches in India. Former Australian cricketer Jason Gillespie asked why Bangladesh couldn’t be afforded the same accommodation India received in the past. He was trolled and had to take the post down.

This use of such power can be seen through an analytical framework. What India is losing in the cricketing domain is not power, but authority. Authority is power plus legitimacy. Legitimacy disarms others, making them willing to accept your leadership without coercion. The US, after the fall of the USSR, had both. What accelerated American decline was the erosion of legitimacy through unilateral actions that alienated allies.

India’s cricket dominance is following a similar trajectory. Though other cricketing boards might not make it explicit, they would be considering forming their own plurilaterals to counter Indian power.

Perhaps every great power behaves in ways we won’t like, but different powers misbehave differently. China’s mercantilism is extractive; America’s hegemony was often militaristic; India’s cricket dominance is proving to be neighbourhood-obsessed.

The point is that India is still not a great power in domains that matter far more than cricket—hard security, economic might, technological capacity. We still have time to shape how India will turn out. The current cricket conduct is worrying precisely because it is a preview. If India deploys neighbourhood obsession, thin-skinned nationalism, and institutional capture at this scale for a sport, imagine what happens when the stakes are genuinely high. This cannot be said about the US or China anymore. Their characters are mostly formed. India’s is still being written. Therein lies some hope.

---

HomeWork

Reading and listening recommendations on public policy matters

1. [Puliyabaazi] In our next episode, we discuss Prosperiti’s analysis of the four new labour codes in India, and the importance of flexibility in working hour regulations.

2. [Paper] This is a foundational paper on reservations by Madhav Khosla and pratap bhanu mehta, which advances a theory of caste formalism. These are the hero lines:

   One of our principal aims is to advance a theoretical argument about what we

   term caste formalism. This formalism manifests in at least two ways. First, caste operates as a rigid classification scheme, presuming causal effects on individuals without further specification. This detachment of group membership from its causal impact immobilizes or prejudges the effects of caste on individuals. It treats caste as the dominant explanatory variable for many different conditions—to which various castes can appeal—overshadowing other factors. The most visible manifestation of this is that mere lack of proportional representation is increasingly a measure of backwardness, and both the under-and-overrepresentation of a caste group provokes suspicion. In this way, the turn to caste membership in and of itself has universalized a structural theory of caste-based discrimination. There is neither place for individual experiences in the causal narrative that explains social outcomes nor for group experiences that may not be reducible to caste.

   Secondly, purely formal terms define the policy instrument of affirmative

   action, namely achieving numerical quotas. One measures the success of the

   policy by the attainment of these quotas. If a caste group secures a designated

   number of seats, the mechanism is a success. Just as the lack of representation is a signifier of backwardness, the achievement of representation provides the suitable remedy. This formalism reduces the ideal of substantive equality to a principle of equal representation on the basis of caste for all caste groups, offering representation as the answer to diverse concerns such as inequality, backwardness, and discrimination. The current framework no longer adequately attends to the mechanisms that produce exclusion—specifically, the link between discrimination and inequality—and it collapses the distinctions between different kinds of harms and disadvantages that citizens might suffer. In doing so, it not only bears an arbitrary character but also risks obscuring persistent practices of discrimination, even within Dalit communities, by prioritizing representation over other forms of redress.

3. [Blog] An Analysis of the India-EU Technology Action Agenda, with recent developments serving as the backdrop.

---