India’s data centres look to the salty seas

In today's episode, we'll break down two important stories. First, we talk about India's data centers looking to the salty seas and then we'll talk about the 10 rupee battery with a 50 crore rupee problem. Welcome back to the Daily Brief by Zerotha, where we cut through the noise to help you understand what's actually happening in the most important stories from business and markets. If you are listening to this on your commute, on a walk or at the gym, you can also find the Daily Brief as an audio podcast on Spotify, Apple podcasts or wherever you listen to your podcasts. If you prefer reading, check out the newsletter using the link in the description. I am your host Akshara and today is Monday, 29 June. Coming to the first story. Recently, the Hindu business line ran an interesting story that caught our eye. India's data center operators are looking to find new innovative ways to cool their extremely thirsty buildings. Reliance, in fact, is looking to use sea water to do so for its AI data center in Jamnagar. Now, it's well known that data centers consume a significant amount of water in order to cool servers and a large, high-priced AI facility can guzzle millions of liters a day. For instance, Google's own reporting says that in 2024, its data centers consume 31 billion liters of water, which is enough to irrigate 54 golf courses or even supply to a small town. India, too, is now in the middle of the most aggressive data center built out in its history. Install capacity has tripled from about 0.5 gigawatt in 2020 to roughly 1.5 gigawatt by 2025 and could cross 6 gigawatt by 2030. Now India doesn't necessarily face a permanent freshwater shortage, but this year's monsoon is a reminder of how fragile the supply can be. The IMB has focused below normal rainfall under el-Nenio conditions and groundwater tables in cities like Mumbai and Chennai have been dropping steadily. Maintaining the freshwater reserves that farms, cities and industries depend on is hard enough without adding massive data centers to the queue. The ocean, on the other hand, offers a practically infinite resource, but even without data centers in the picture, turning sea water into something a data center can actually use is harder than it sounds. Few countries have managed to make it an industrial scale process and we were curious to know why that is. So let's start with how data center cooling works. Thousands of servers packed into a single facility generate enormous heat. Left uncooled chips could fail within us and even minutes. The most common solution is evaporative cooling. Here cold water is spun through pipes near the servers, absorbs the heat and is pushed through cooling towers where it evaporates. The evaporation carries the heat away but consumes the water in the process. Now assume this process uses sea water. As the water evaporates, any dissolved minerals or salts left behind get progressively concentrated in the remaining water. And over time, those minerals form hard, chalky crusts on pipes and heat exchangers clogging the system. And bacteria and algae thrive in warm mineral-rich water forming slimy biofilms that further choke the equipment. In essence, sea water would wreck a conventional cooling system almost immediately and the salt would corrode metal components, the minerals would scale up the pipes and marine microorganisms would foul the surfaces. So data centers need water that's clean enough to cycle through the system without leaving destructive residues. And that means fresh water from rivers, reservoirs, municipal supply or groundwater, which is precisely the resource India can't afford to divert in large quantities during poor monsoons. So the alternative is to take the infinite supply, sitting off shore and strip the salt out of it. Now there are three main industrial techniques for sea water desalination, each with its own logic. So the oldest approach is multi-stage flash distillation or MSF. The technique of MSF as a sequence of sealed chambers each held at a progressively lower pressure. Now sea water is heated and then fed into the first chamber and when it enters a low pressure environment, a fraction of it flashes into steam. That steam condenses on cold pipes and drips down as pure fresh water which then flows into a separate channel. Now MSF is reliable and proven but brutally energy intensive. It takes close to 120 kWh of heat to produce 1000 liters of fresh water plus additional electricity for pumping and it also produces the most expensive water of the three methods. Now the second method is multi-effect distillation or MED and it's a more efficient version of the same principle. So instead of flashing, it uses a staircase of evaporation effects. Steam produced in one stage heats a thin film of sea water in the next, which boils and produces more steam and so on down the chain. Because MED transfers heat more efficiently through these thin films, it uses much less energy than MSF. Now both MSF and MED produce extremely pure water which is valuable for specialized industrial applications. But while less expensive than MSF, MED still comes at a steep cost. Now the third method which also dominates modern large scale decalination is reverse osmosis or RO. The same technology used in your home water purifier. So here sea water is posed through semi-permeable membranes at extremely high pressure and the membranes have forced small enough to let water molecules through but block the larger salt particles and unlike MED and MSF RO requires no boiling. But the trade off is that while RO filters salt out, other impurities may require separate tension. The membranes demand rigorous pretreatment of the incoming sea water like chlorine to kill bacteria, coagulants to remove suspended particles, acids to prevent mineral crusts and then de-clordination before the water hits the membrane since chlorine damages the membrane material itself. But RO's ACE which also enables its large scale application is its energy bill. Because it doesn't boil anything, it only needs about 4 to 5 kWh of electricity per cubic meter just for the high pressure pumps that force water through the membrane. And that's an order of magnitude less energy than MSF. Additionally, over the years, the RO process itself has become more efficient with time, making the unit economics of an industrialized process better. But even with RO's efficiency gains, desalination remains significantly more expensive than drawing water from a river or an aquifer. Producing 1 cubic meter or 1000 liters of desalinated sea water typically costs between 0.5 to 1.5 dollars depending on energy prices, plant scale and local conditions. Natural sea water, where available, can cost as little as 0.1 dollars per cubic meter. So this expense breaks down into two buckets. Up front capital is large. Building a factory scale RO plant with intake pipes, pretreatment systems, membrane arrays, energy recovery devices and brine outfall infrastructure costs 1,000 to 2,500 dollars per cubic meter per cubic meter. A plant producing 1 lakh cubic meters a day might require 100 million dollars to 250 million dollars just to build. And then there's operating cost dominated by electricity, which accounts for 35 to 45% of ongoing expenses. So in the world, the countries that have pushed the desalination frontier the most are also the nations that were desperately starved of freshwater. The UAE, Israel, Saudi Arabia and other countries in West Asia. Now Israel, a country where 70% of the landmass is desert, now gets most of its drinking water from desalination. Israel has five large RO plants along the Mediterranean coast, which together produce about 600 million cubic meters a year and supply 80% of Israel's urban household water needs. In fact, Israel has such a huge surplus of desalinated water that for the first time anywhere, it's now trying to refill a freshwater lake with the same. Saudi Arabia operates on an even larger scale. Their Al Jubeil desalination plant is one of the largest in the world, converting over 1.4 million cubic meters of water a day. And it's also not entirely a coincidence that it also has the world's largest oil capacity. Oil refineries already possess ports, power generation, pipelines and industrial utilities that can also support desalination. Now India's experience with desalination is real if less celebrated. Most of our capacity is concentrated in one state, Tamil Nadu. Our first desalination plant was built at Minjur Tamil Nadu and could deliver 100 million liters a day or MLD. After a near catastrophic water crisis in 2019, Tamil Nadu has doubled down on desalination capacity and the town of Namali has two plants which can collectively deliver 250 MLD. A fourth planted perot will add another 400 million liters per day by early 2027. And once operational, it will be one of Asia's largest seawater arow plants and desalinated water will supply roughly half of Chennai's drinking water. Meanwhile, in Kalpaka, in Tamil Nadu, the Baba Atomic Research Centre has been running a hybrid nuclear desalination plant since the early 2000s, coupling both MSF and arow processes to the Madras Atomic Power Station to produce 6.3 million liters a day. BARC has also deployed smaller solar powered arow units in remote villages and emergency barge mounted plants for disaster relief after the 2004 tsunami. Now Reliance has been in the desalination game for longer. They have been operating captive MED and arow plants at their Jamnagar Refinery Complex since the 2000s. built with Israel's IDE technologies and Vietek Wabakh. The ultra-finery needed ultra-poor water for processing. Now, the same infrastructure extends to the data center campus next door. So here's where the story gets interesting for India specifically. The single biggest cost in running a desalination plant is electricity. And those costs used to be elevated with coal power. But two of India's largest data center builders, Reliance and Adani happened to also be the country's two biggest renewable energy developers. So Reliance is building a massive integrated renewable energy ecosystem in Gujarat. Solar PV manufacturing in Jamnagar, battery storage systems and multi-gigawatt solar farms in Kutch that the company says will be among the world's lowest cost sources of round the clock green electricity. Using your own solar power at rock bottom cost to desalinate sea water for your own data center improves the economic significantly compared to a municipal utility buying electricity at grid rates to run a desalination plant for the city. Adani has a similar structural advantage. So Adani connects the group's data center joint venture is explicitly building renewable powered facilities and Adani's position in power generation, solar manufacturing and transmission gives it the same kind of vertically integrated cost structure. So both companies can in principle co-locate data centers, solar farms and desalination plants on the coast turning sunlight into compute with sea water as the coolant. A recent study of a large RO plant in the Mediterranean backs the economics up. So powering a desalination facility with a hybrid mix of solar wind and grid electricity cut the cost of water production by about 20% compared to relying purely on the grid. The renewables setup also acted as a hedge against fossil fuel price volatility. And on top of the cost savings, the hybrid plant cuts its carbon footprint by about 35%. Now desalination won't solve India's water problems. It's expensive. It only works for coastal regions and dumping the brine byproduct out of desalination back into the ocean damages marine ecosystems if it isn't managed carefully. But for data centers which need massive amounts of water, unpredictable uninterrupted supply and which are increasingly being cited on India's coastline anyway, it offers independence from the freshwater grid at a time when that grid is under growing pressure. But even then, when it comes to the data center build out, a desalination strategy only makes sense in a few places. Mumbai and Chennai, two of India's biggest cities and also next to the sea, will benefit from this. Jamnagar also has a coastal town. But for inland areas like Hyderabad and Bengaluru, the water problem may have to be solved differently. And what also remains to be answered is whether reliance is vertically integrated model at Jamnagar becomes the template for India's data center build out or remains an exception that only large conglomerates can afford. Coming to the second story. Panasonic has been making batteries in India since 1972 filling our TV remotes and wall clocks with double A and triple A cells. But last week it said it might have to stop. Not because demand collapsed or there was immense competition, but because of a recycling rule. So we told business standard that its only Indian battery plant in Pithampur, Madhya Pradesh may become unviable under the country's battery waste management rules. Now by all means this is no small operation. The plant employs 283 people, makes 54 crore batteries a year and holds about a fifth of India's Rs 4000 crore dry cell market share. Last year it earned a net profit of Rs 3.4 and crore. But the compliance bill it's now staring at could be around 8 times that. Now Panasonic's own auditor has also flagged the issue, qualifying the F-826 numbers because the company hasn't set aside any money for what the rules will eventually cost. So when the auditor gets nervous and issues a qualified opinion, something is real. But what's the big deal about these rules? Now we've covered before about how India's e-waste rules went from an idea to a real enforced market. And batteries followed a similar script anchored by a concept called extended producer responsibility or EPR. So in simple terms what EPR meant was that if you made money selling something, you're also on the hook for what happens to it when it dies. Under the battery waste management rules of 2022, every producer has to make sure a slice of what it once sold gets collected and recycled. For dry cells this year, that slice is 50% by weight of what it sold three years ago. Next year it'll clamp to 60% and the year after 70%. So if you sold 100 kgs of dry cells in FY23, you're responsible for 50 kgs of them to be recycled by the end of this year. So how does a producer actually pull this off? There are two ways. You can either do it yourself by building the collection network, gathering the dead batteries and getting them to a recycler, or you can let someone else do all of that and pay them for it. So the second rule runs on something called an EPR certificate. So think of it like a carbon credit. A registered recycler does the actual recycling. The government's portal issues a certificate for the quantity processed and the producer buys that certificate to prove it hits target. The recycler gets a fair sum. The producer takes responsibility for their waste and the government manages to get the circular economy of batteries up and running. Now what a producer can't do is nothing. Miss the target and you pay a penalty called environmental compensation and what's more, paying the penalty doesn't erase the obligation because the shortfall will simply roll over to the next year. This is the very system pulling India's lead acid and lithium batteries into the formal economy. It sounds simple enough, but the trouble starts the moment you ask what kind of battery is subject to this rule. Now when we say battery, we're using one word for three completely different businesses. A lead acid car battery for instance is heavy, expensive and full of lead worth real money. When it dies, a mechanic or scrap dealer is happy to take it because there's cash inside. Amararaja, one of India's biggest lead acid makers, spent Rs.700 crore building its own recycling plant precisely because recovered lead is cheaper than buying it fresh. For lead acid recycling is much less a burden than it is a raw material procurement strategy. But a lithium battery, the kind in your phone or an EV is newer and harder to recycle, but it's packed with critical minerals like lithium, cobalt and nickel. India imports almost all of those much of it from China. So the government badly wants them recovered at home. So there's a whole apparatus pushing this along, including rupees 1500 crore incentive scheme for recyclers, custom duty waived on batteries scrap and the national critical minerals mission treating recycling as a strategic priority. And then there's the humble dry cell with zinc and manganese worth pennies. So the problem lies in the fact that the EPR road street all three the same way. So to actually comply a producer by certificates and the government has fixed what dose cost metal by metal and the price very strongly by chemistry. While it's 18 rupees a kilo for lead, it's a whopping 2400 rupees for lithium. Fair enough on paper because trickier costlier chemistries should cost more to handle for lithium. A high number makes sense so that it pulls in more players in the recycling industry because the recovered metal is genuinely valuable and the government is pouring money in behind it. We can't get enough of lithium and we need more lithium recyclers. But zinc, the main metal in a dry cell with a much simpler chemistry sits right at the top of that ladder at lithium's level at rupees 2400 per kg. And there's one more screw to turn from 2027 dry cell makers will also have to start putting recycled material back into their new batteries beginning at 5% and rising over time. The catch, parasonic says is that recycled zinc carbon materials today isn't pure enough to reuse and as per them, it's a mandate to use material that barely exists yet. Number four, any of this matters, you have to physically get the batteries back and dry cells are almost engineered to vanish. See, a dead car battery goes back through a dealer. Even a dead phone has resale and scrap value, but a dead AA battery which is much smaller than a car battery or a phone battery usually ends up in the kitchen bin with the vegetable peels and inevitably to a landfill. No dealer, no deposit, no reason for anyone to fish it out. Even India's enormous network of Kabadiwalas was so good at grabbing value out of waste that they outbid the formal recyclers, won't touch a dry cell just because there's nothing in it for them. Now India says somewhere between 2.2 and 3 billion dry cells a year. Even the companies running pilot collection drivers are recovering only about 20% of them, but the rules ask for 50% starting now. So the industry keeps pointing at the countries that did this properly. Switzerland and Belgium reached 70% collection, but that took 10 to 15 years of patiently building the bins, the habits and the logistics. India set a 50% target in year 1 with none of that in place. As the head of Nippo put it, these systems are usually phased in over 6 or 7 years, not switched on overnight. So what this means is that Panasonic's problem isn't just recycling, but also the reverse logistics. Nobody collects dry cells, so recyclers have almost nothing to process because of which very few EPR certificates get created. Now with few certificates, but also plenty of producers who legally need them, including Panasonic, the price gets bit up. And since the flow price for those certificates is set so high for zinc, it started out expensive anyway, scarce and costly by design. So Virgin Zinc cost around Rs. 300 a kilo. The regulated unit cost of zinc recycling, Panasonic says, runs about 240% higher than that. So you're paying a steep premium to recover a metal you could simply buy cheap. The recycling uses money before you've even counted the cost of collecting anything. For Panasonic specifically, if it had to meet the full 50% target by buying certificates, it estimates the cost would reach around Rs. 50 crore. Against a Rs. 3.49 crore profit, that wipes out the business several times over. And the penalty rises 10% every year, so the squeeze tightens on its own. Now it would be easy to read all this as one foreign company complaining. But it isn't. Ever-ready, nippo, and uracel, all well-known household names have all raised the same alarm. Now they're not asking to scrap the rules. Instead, they want three fixes. Lower the first-year target, phase it in over years instead of demanding 50% on day one, and reprise the compliance cost for zinc to reflect what a dry cell is actually worth. Now that's not what lithium is worth, but that wouldn't attract enough recyclers, so that's another problem. Some are floating ideas to kickstart collection like handing out free batteries in exchange for used ones or pulling a single collection network across the whole industry. The dry cell is the worst case, but the flaw underneath it runs through the whole framework. The flaw of the rule that assumes a recycling market already exists, which clearly doesn't, especially EVs. So from 2027-28, makers of lithium batteries will have to use a minimum share of domestically recycled lithium, cobalt, and dinkle in their new cells. Sounds great. Except India today formally recycles less than three percent of its lithium ion batteries, and most of the EV batteries sold so far are still sitting in cars on the road, and they won't die for years yet. So the rule will demand recycled content that physically barely exists, same shape as the dry cell problem. The mandate shows up before the material does. But the difference is that with lithium, everyone actually wants to solve it. The metals are valuable. India needs them for its own security, and the state is pouring money in to build the recycling industry ahead of the mandate. Lithium can grow into the rule, but dry cells get the rule without the help. Now, when you strip away the noise, you can see that both sides actually agree on the principle. Battery shouldn't end up in landfills, producers, not taxpayers, should pay for the cleanup. EPR is a good idea, and India's version of it does seem to work for the battery's worth working on. The disagreement is narrower than it looks, whether you can demand a mature recycling market before you've actually built one, and whether a rule priced for lithium should land unchanged on a metal worth of fraction as much. How the government answers that will decide a lot more than Panasonic's fate, because if a rule designed to build a circular economy ends up closing the factories at its low-value edge, the Panasonic plant won't be the only loss for India. We'll also be left with the same two and a half billion dead batteries a year without those who we need to come and collect them. Now coming to the tidbits. The Bureau of Energy Efficiency or BEE has proposed a voluntary one to five star rating system for electric two-wheelers, similar to ratings for appliances. The labels will help buyers compare energy efficiency while encouraging manufacturers to build more efficient e-scoters. Coming to the next tidbit. Adani Group plans to enter the nuclear power business through Adani-atomic Energy, targeting 10 gigawatt of capacity by 2035. The company also announced a Rupees 2 lakh crore power expansion plan, higher data center capacity, and new investments in hydro power and digital infrastructure. Coming to the final tidbit. Biocon, Ziders, Dr. Reddy's, and other Indian drug makers are preparing to launch bio-similar versions of Merck's Cancer Drug Key Truda as its patents begin expiring from 2028. Lower cost alternatives could cut treatment prices by 50 to 80% making the therapy accessible to many more cancer patients. That's all the news I have for you. Thank you so much for watching and see you in the next one. [BLANK_AUDIO]