In Part I of this series, we argued that the drone industry is maturing into a technology stack. The market is no longer being shaped by companies trying to build an entire aircraft from scratch. It is increasingly being defined by specialised layers, each solving a distinct engineering problem. Batteries, propulsion, avionics, communications and onboard electronics are becoming products in their own right. That was the thesis behind our investment in ZeroDrag. Reliable electronics are no longer a feature of the drone. They are becoming infrastructure for the industry itself.

Every infrastructure industry evolves the same way. Once one layer becomes reliable, attention shifts to the next constraint. Better semiconductors made cloud computing possible. Cloud computing accelerated artificial intelligence. Artificial intelligence is now driving another generation of investment into compute and power infrastructure. None of these layers replaced the one beneath them. They made it more valuable.

The drone industry is entering a similar transition: reliable electronics have made autonomous flight commercially viable; the next challenge is enabling these systems to understand the environment they operate in.

We still call them drones because flight is their defining characteristic. Over time, that language will become increasingly inadequate; the engineering bottlenecks shaping the next decade are no longer unique to aviation, they're robotics problems.

GPS Built the First Generation of Commercial Drones

The commercial drone industry benefited from solving one of robotics' hardest problems before it had to solve it itself.

Every autonomous machine needs to answer a deceptively simple question before it can perform any task: Where am I? For the first generation of drones, GPS answered that question with remarkable reliability. Satellite navigation transformed localisation into a largely solved problem, allowing manufacturers to focus on everything else. Better batteries increased endurance. Better motors improved payload capacity. Better airframes reduced cost and improved reliability. Navigation itself rarely became the limiting factor because the environments drones operated in were remarkably forgiving.

That first generation of commercial applications reflected this reality. Aerial mapping, surveying, photogrammetry and infrastructure inspection all happened in open environments with uninterrupted visibility to the sky. The aircraft did not need to understand the world around it. It only needed to know its position within it.

GPS was the invisible infrastructure that allowed the industry to develop as quickly as it did. GPS allowed the industry to treat drones as aircraft, and as GPS becomes less reliable, they increasingly have to behave like robots.

The Market Moved Closer to the Ground

The nature of commercial drone operations has changed. The next generation of applications expects drones to work within the environment rather than above it. They inspect transmission corridors running through dense vegetation, monitor industrial plants filled with steel structures, navigate warehouses where satellite signals do not exist, survey mines, inspect bridges from underneath and increasingly support defence operations where GPS disruption is expected rather than exceptional.

The closer an autonomous system moves towards the physical world, the more demanding localisation becomes.

GPS performs exceptionally well when an aircraft has a clear view of the sky. It performs considerably less well in reflective structures, inside enclosed environments or wherever satellite signals are blocked, distorted or deliberately denied. Even where GPS remains available, positional error that is acceptable during long-range mapping becomes unacceptable when operating centimetres away from infrastructure. A two-metre deviation is largely irrelevant when surveying hundreds of acres from altitude. It becomes mission-critical when flying alongside a conductor, over a plantation canopy or through an industrial facility.

It describes many of the highest-value commercial applications for autonomous flight.

Localisation Has Become the Bottleneck

The industry's response has largely been incremental. Higher-precision GPS through RTK corrections, improved inertial navigation, LiDAR and increasingly sophisticated sensor fusion have all extended what autonomous aircraft can do. These technologies have improved reliability, but they do not fundamentally change the problem.

Localisation in unstructured environments requires more than accurate coordinates. It requires contextual understanding.

An autonomous drone inspecting transmission infrastructure is not simply navigating to a GPS waypoint; it must understand where conductors, towers, vegetation and obstacles exist relative to its own position. A drone operating inside a warehouse has no satellite reference at all. A defence platform operating in a contested environment cannot assume GPS will remain available throughout a mission. In each of these cases, the challenge is no longer estimating position relative to satellites overhead; it's estimating position relative to the environment itself.

That distinction changes the nature of autonomy. For much of the last decade, localisation depended primarily on external infrastructure. Increasingly, it is becoming a function of onboard perception.

This is also where the language begins to matter. We still describe these systems as drones because flight is their most obvious characteristic. Increasingly, however, they are behaving like robots, with their competitive edge depending on how well they understand the world around them.

Why Agriculture Matters More Than People Think

Agriculture is often presented as the primary market for commercial drones.

We think it is better understood as the proving ground for the next generation of autonomous systems.

Few commercial environments combine as many localisation challenges within such a small operating envelope. Dense canopy reduces satellite visibility. Uneven terrain changes altitude continuously. Dust, changing light and dynamic vegetation create an environment that rarely looks identical from one flight to the next. Precision agricultural operations demand centimetre-level positioning while operating close to the ground, where satellite navigation becomes increasingly susceptible to multipath effects and vertical error amplification. The receiver may continue reporting a reliable GPS fix even as its true localisation accuracy deteriorates.

Solving autonomy under these conditions has implications far beyond agriculture.

The same localisation capabilities apply to infrastructure inspection, industrial facilities, warehouses, mining, logistics, utilities, ports and ISR (Intelligence, Surveillance, and Reconnaissance). These markets appear unrelated, yet they all require autonomous systems capable of operating close to physical infrastructure without depending entirely on satellite navigation.

Agriculture matters because it compresses one of robotics' hardest engineering problems into an environment where the commercial value is already evident.

The Economics of Autonomy Have Changed

For years, vision-based localisation remained technically promising but commercially impractical. Cameras were inexpensive, but the computational resources required to process visual information in real time imposed unacceptable penalties on cost, power consumption and weight.

That equation has changed.

Camera hardware has become commoditised. Edge AI processors now deliver significantly greater performance within the power budgets robots can accommodate. At the same time, perception algorithms have become fundamentally more efficient, extracting more capability from limited compute and enabling sophisticated vision models to run entirely onboard. Running perception at the edge has become commercially viable rather than experimentally interesting. Vision also offers something that satellite navigation never could. It does not simply estimate where a machine is. It enables the machine to interpret what exists around it.

This is an important distinction. Coordinates tell a drone where it should be; perception tells it what is actually there.

That shift transforms localisation from a navigation problem into a software problem. Performance increasingly depends on field data, iteration and algorithms rather than individual hardware components. Competitive advantage accumulates through deployed systems learning from increasingly complex environments, creating software moats that are difficult to replicate through manufacturing alone.

The Next Layer of the Robotics Stack

Part I argued that reliable electronics would become foundational infrastructure for autonomous aviation. We continue to believe that.

Reliable avionics remain indispensable because every autonomous robot depends on them. But as one layer matures, value creation moves upwards. The next generation of autonomous systems will not differentiate themselves solely through better airframes or longer flight times. They will differentiate themselves through systems capable of understanding complex physical environments with enough confidence to operate independently.

That shift extends well beyond the drone industry. Every autonomous machine operating close to the physical world faces the same underlying challenge: it must localise itself accurately without assuming perfect connectivity, uninterrupted satellite coverage or structured environments.

Perhaps that's why calling them drones increasingly feels incomplete – a drone describes how the machine moves; a robot describes what the machine is capable of doing. As autonomy moves closer to the physical world, that distinction becomes increasingly important. As autonomy expands across infrastructure, logistics, industrial operations and defence, localisation is becoming foundational infrastructure in its own right.

The first generation of commercial drones was built on GPS. The next generation will be built on systems capable of understanding the world when GPS is no longer enough, or unavailable.

We think this is the next important layer of the autonomy stack. It's a layer we're increasingly convinced will shape how autonomous systems are built over the coming decade. We've made a bet on it.

More on that soon.

Why cooling and humidity suddenly matter

Walk into a pharmaceutical plant or a spice factory in coastal India during the monsoon, and you experience, viscerally, that "comfort cooling" is the wrong frame. The system is not just fighting temperature; it is fighting water in the air, and that latent load forces compressors, coils, and desiccant wheels to work longer and harder than their nameplate capacity suggests.

From a thermodynamics lens, industrial and commercial cooling is always simultaneously handling sensible heat (lowering temperature) and latent heat (removing moisture). In hot‑dry climates most of the energy goes into dropping temperature, while in hot‑humid climates like Mumbai or Chennai up to 30–40% of the cooling work is diverted into dehumidification, dramatically increasing energy consumption and making humidity control a product‑quality and energy‑cost problem rather than a comfort feature.

The market and its incumbents

Industrial cooling and dehumidification today is dominated by three technology families: refrigeration‑based systems (standard vapor‑compression HVAC), solid desiccant systems (rotors and wheels), and liquid desiccant systems, each with distinct physics, cost structures, and use‑cases. Refrigeration is ubiquitous in commercial buildings and data centres, solid desiccants are standard in process‑critical industries needing deep drying, and liquid desiccants sit as a niche but high‑potential bridge technology that promises independent control of temperature and humidity.

In India, this abstract landscape translates into a concrete installed base: industrial dehumidification across pharmaceuticals, textiles, chemicals, and food and beverage alone represents roughly 2.5 billion dollars of equipment in the ground and about 500 million dollars of annual recurring market from replacements and retrofits. This market is roughly split between solid desiccant and refrigeration‑based systems in these segments, with a handful of global and domestic players like Bry‑Air, Munters, Seibu Giken, Advancel/Drycool and others holding the solid‑desiccant share, while a thin layer of integrators imports liquid‑desiccant systems from companies such as DuCool, Advantix, Alfa Laval's Kathabar and emerging US players.

Policy and macro tailwinds

Policy and economics together are shifting the incentives for every kilowatt‑hour of cooling. Rising power tariffs and energy‑intensive manufacturing mean efficiency now shows up directly in plant P&Ls, not just in ESG reports. US and European climate packages explicitly reward high‑efficiency HVAC and low‑carbon industrial processes, while India has committed to ~1 gigatonne of emissions reduction by 2030 and is pushing energy‑efficient technologies through programs like the Bureau of Energy Efficiency and FLCTD (Facility for Low Carbon Technology Deployment).

At the same time, global capital is moving decisively: BloombergNEF estimates that, even after current climate investments, the world still needs ~$3 trillion of additional annual investment to meet net‑zero targets, and a meaningful slice of that will flow into technologies that reduce industrial and building‑sector energy use rather than only generating more clean power. For cooling, this translates into very simple buying logic: if a system can cut energy use per tonne of cooling or tap waste heat, it can usually clear an industrial customer's hurdle rate even before you price in carbon.

How the cooling stack is built

Inside factories and large commercial buildings, the cooling stack is built from standard blocks - a generation system (chillers or packaged units), air handling units, ventilation/dehumidification equipment, heat rejection (condensers or cooling towers), controls, and distribution through ducts or process piping. In commercial settings, this often means rooftop units, standard DOAS systems, medium‑grade filters, and air‑cooled condensers focused on "comfort," while in industrial settings it means centrifugals or screw chillers, desiccant DOAS, stainless steel AHUs, medical‑grade filtration, and water‑cooled towers built for 24/7 process reliability.

Three architectures are used to remove moisture at scale:

Refrigeration Systems

Refrigeration systems use a vapor‑compression cycle - standard chillers and rooftop units - to cool air below its dew point on cold coils, condense out water, and then reheat the now‑dry air to the required supply temperature. They are mature, widely available, and excellent at dropping temperature quickly, which makes them the default choice in commercial buildings and data centers, but they become inefficient and hit a hard limit when pushed to very low dew points because coils start to ice and energy use spikes with repeated cool‑and‑reheat cycles.

Solid Desiccant Wheels

Solid desiccant systems use a slowly rotating wheel coated with porous materials like silica gel or zeolites that adsorb moisture directly from the air, then regenerate using high‑temperature air in a separate zone. This architecture delivers extremely low dew points and relative humidity, which is why it is standard in pharmaceuticals, specialty chemicals, food processing, defense storage, and power‑plant turbine lay‑up, but it demands 120–150°C regeneration heat and adds significant thermal load that must be cooled again, driving up total energy consumption.

Liquid Desiccant Systems

Liquid desiccant systems pass air over or through a hygroscopic salt solution that absorbs moisture at the molecular level, then use moderate‑temperature hot water to strip that moisture out in a regenerator and recirculate the dried liquid. They can decouple temperature and humidity control, use low‑grade waste heat, and avoid the over‑cool‑and‑reheat penalty of pure refrigeration, yet conventional chemistries have been corrosive or toxic, which has limited adoption to niche, high‑spec projects despite strong thermodynamic and energy‑efficiency advantages.

Together, these three architectures define the engineering landscape into which any new cooling technology must fit: the familiarity and ubiquity of refrigeration, the performance of solid desiccants, and the efficiency promise of liquid systems.

The structural trade‑offs inside today's stack

Industrial buyers are effectively choosing between three imperfect options. Refrigeration systems are mature and easy to deploy, but they struggle to deliver deep drying and end up over‑cooling and reheating air, wasting energy and hitting coil‑freezing limits below roughly 4°C dew point. Solid desiccants deliver the humidity performance that process‑critical industries need, yet they pay for it with 120–150°C regeneration temperatures, additional cooling stages and elevated electricity consumption in continuous operation.

Traditional liquid desiccant systems offer strong thermodynamic advantages by decoupling temperature and humidity and using low‑grade heat, but they rely on corrosive or toxic salts like lithium chloride and lithium bromide, which drive capex up through exotic materials and create safety concerns about chemical carryover into supply air. The result is a three‑way trade‑off: high‑opex deep‑drying, high‑opex comfort cooling, or high‑capex imported efficiency, with little domestically tailored innovation for hot, humid Indian conditions.

What the market actually looks like

On the ground, the demand side is anything but niche. In India alone, more than 10,500 pharmaceutical plants, thousands of large textile and garment factories, tens of thousands of chemical plants, and around 2,000 food and beverage facilities embed dehumidification into their core processes, making low‑humidity environments mission‑critical for tablet production, powder handling, lamination, seed storage, defense mothballing, and data‑center reliability. Many of these plants already operate solid desiccant systems and precision HVAC, yet they face steadily rising electricity tariffs and tighter product‑quality and regulatory standards.

On the supply side, the domestic market is well served in solid desiccants but has almost no homegrown liquid‑desiccant OEMs; local integrators tend to import systems from Israel, Europe, or the US, making capex high and support fragmented. Globally, liquid‑desiccant innovation is active (companies like Blue Frontier, Mojave, 7AC and others are funded or acquired for their efficiency and grid‑shifting capabilities) but they primarily target commercial buildings in Western markets, carry high equipment complexity, and have not yet localized for Indian industrial conditions.

The gap we see

Within this context, the gap is clear. Indian industrial customers increasingly need systems that can deliver solid‑desiccant‑level humidity performance with refrigeration‑like or better energy efficiency, while avoiding the corrosion and toxicity issues of standard liquid desiccant chemistries. They also need equipment that slots into existing AHU and process‑air footprints, interfaces cleanly with chilled‑water and hot‑water loops, and can tap low‑grade waste heat rather than demanding bespoke 120°C steam or electric heaters.

This combination leaves a wide, strategically important space for technologies that rethink liquid desiccants around non‑corrosive, non‑toxic chemistries and India‑first engineering.

Who will win

The companies that win in this segment will be those that treat industrial cooling as an energy‑transition problem, not an HVAC line item. On the technology side, they will build liquid‑desiccant systems that regenerate using low‑temperature heat compatible with waste‑heat or simple hot‑water loops, integrate cleanly with existing chilled‑water infrastructure and deliver cooling and dehumidification in a single, tightly coupled step. Instead of relying on highly corrosive chemistries and exotic alloys, they will use benign formulations and commodity materials that keep capex and long‑term maintenance costs in check.

Architecturally, these systems will show up as modular, retrofit‑friendly components - advanced heat‑and‑mass‑exchange modules that can slot into or alongside standard AHUs - turning conventional air‑handling boxes into high‑precision, low‑energy air treatment systems without demanding greenfield plant designs. Commercially, winning companies will follow phased GTM: starting with India's industrial dehumidification segments where the problem and ROI are well understood (pharma, textiles, chemicals, food), building credibility through pilots and ESCO partnerships, then expanding into global DOAS markets and finally into commercial AHU/VRF systems once field performance and cost curves are proven.

These companies will be led by engineering‑strong founding teams comfortable with desiccant science, fluid dynamics and thermal systems, and will consciously supplement that with sales, GTM and OEM‑partnership capability, because selling a new category of cooling technology into conservative industrial environments requires both technical depth and narrative skill. In effect, they will be building a new layer of industrial infrastructure for hot, humid markets, one where moving heat and moisture intelligently becomes as strategic as generating clean power.

When people think about progress in LEDs or power electronics, they usually look at the obvious levers: brighter chips, tighter drivers, clever optics. But if you trace failure modes and warranty claims backwards, you keep ending up in a less glamorous place: the metal‑core PCB, and specifically the thin dielectric layer that decides whether heat ever leaves the system at all.

In our previous piece, The thermal bottleneck inside modern lighting, we walked through the physics, policy tailwinds, and market sizing behind this problem. This note is the footnote to that thesis: here is the company we chose to build on top of it.

Helionis Labs is our bet on this layer. This is why.

What Helionis is Building

Helionis is an IMS (Insulated Metal Substrate) design and manufacturing company focused on solving the thermal bottleneck at its root cause.

Conventional MCPCBs rely on laminated dielectric sheets and copper etching. This limits thermal performance, creates material waste, and effectively blocks true additive manufacturing. Helionis replaces this with an in‑house printed, ceramic‑filled dielectric and silver‑based additive circuit printing. That combination enables thinner insulation, better heat flow, higher yield, and fewer process steps.

In practice, Helionis does three things differently:

• Designs the dielectric in‑house. Instead of buying a fixed laminate, they formulate a composite of polymer matrix, ceramic fillers, and adhesion promoters tuned to aluminium.
• Prints it directly on aluminium. Screen printing allows them to hold dielectric thickness around 50–75 µm versus the 100–200 µm common in standard MCPCBs.
• Prints the circuit additively. Silver inks survive 300–400°C sintering without oxidising, enabling clean traces without copper etching.

The result is an IMS board that is materially more thermally efficient than incumbent MCPCBs, with a shorter heat path and a process that can, in principle, scale inside a factory rather than a lab.

Why This Approach Matters

As we wrote in the pre‑read, the substrate market is barbelled.

At one end, global majors supply high‑performance IMS and thermal PCBs with advanced ceramic‑polymer dielectrics and embedded copper inlays, often reaching 3 W/m·K and above. These are excellent products but are priced for high‑margin, lower‑volume programmes and typically arrive in India as imports.

At the other end, low‑cost MCPCB suppliers use older plastic or epoxy dielectrics. Indian PCB houses assemble aluminium‑backed boards for LED OEMs but rely on imported laminates and do not own substrate chemistry.

Helionis is aiming squarely at that missing middle: a domestically manufactured, high‑performance substrate platform for mainstream wattages, starting with LEDs and expanding into power electronics.

How Helionis Fits India's Supply‑Chain Moment

Another part of our conviction comes from how Helionis sources and manufactures. The company uses commodity raw materials (aluminium, ceramics, polymers, silver) and builds value in the formulation and process, not in fragile or geo-exposed inputs.

We believe this innovation is both a strong value proposition and relatively de‑risked from supply‑chain shocks, because:

• The inputs are commodity materials,
• The process is designed to run in‑house, and
• The company's strategy is to keep core manufacturing under its own roof.

In a world where dielectric laminates are often imported and subject to currency moves, shipping timelines, and geopolitical risk, this matters.

Where Helionis started, and Where It Can Go

Helionis is deliberately starting where thermal pain is visible, and switching costs are manageable: LEDs. The first customers are likely to be Tier‑1 lighting OEMs (street lights, industrial, floodlighting) and automotive suppliers (DRLs, headlamps, exterior modules), as well as fragmented MSME manufacturers who can move faster.

The initial go‑to‑market has three layers:

• Core: IMS boards as components in existing supply chains, replacing MCPCBs inside familiar form factors.
• Support: a small line of high‑power LED luminaires to demonstrate system‑level advantages and act as live reference designs.
• Future: extension into EV chargers, motor controllers, drones, and wide‑band‑gap power systems where thermal dissipation requirements are high and markets are large.

LEDs are a beachhead. The physics that Helionis is engineering against - heat density at the substrate - repeats in each of those adjacent verticals.

The Team We Are Backing

We back teams where the founding skill set matches the bottleneck. In Helionis, that bottleneck is part materials science, part manufacturing engineering.

The founders bring:

• deep experience with LED systems, glass‑based IMS platforms, and high‑power luminaires, and
• global plant operations and reliability experience in chemical manufacturing.

With this combination, Helionis can design the dielectric, build the line, and speak credibly to OEM engineering teams about both lab data and production realities.

Why We Invested

We invested because the company is solving a real industrial constraint at the right abstraction level. The market already has good chips, good drivers, and good system integration, but the board remains a thermal choke point in many high-power applications. Helionis is not trying to win on cosmetic differentiation; it is trying to change the thermal behaviour of the board itself, which is where a meaningful part of the value is trapped.

The other reason is that the business fits how we think the energy transition will be built in India. Transition VC backs engineering-led companies that work on how energy is produced, stored, moved, and consumed. Helionis belongs in that frame because better thermal substrates are the enabling infrastructure for higher power density, longer life, and lower system cost.

The energy angle is also important because heat is wasted energy. Every degree of unnecessary thermal rise forces the system to spend more on cooling, derating, oversizing, or replacement. By improving thermal transfer at the substrate, Helionis is effectively reducing avoidable energy loss inside the hardware stack, which is exactly the kind of physical efficiency gain the energy transition depends on.

Helionis also has the right supply-chain logic for India. Its inputs are commodity materials, its process is designed to run in-house, and its value is created through formulation and manufacturing control rather than imported laminates or scarce inputs.

For that reason, Helionis is not just a materials company or a board house. It is a manufacturing company positioned inside the energy stack, starting with LEDs and extending into power electronics, where thermal management is a constraint on performance and scale. The investment is a bet that this layer will matter more as India localises hardware, raises power density, and pushes more of its industrial energy demand through compact, thermally constrained systems.

India's LED and power‑electronics stack is running into a hard physical ceiling: every generation squeezes more power into smaller footprints, but the thermal "pipe" inside the board has not kept up. Data from lighting deployments, market sizing, and materials science all point to the same conclusion: the dielectric layer in metal‑core PCBs is turning into a strategic bottleneck, and whoever solves it at Indian cost structures will sit in the flow of a compounding market.

This is about that hidden layer - the thin dielectric inside a metal‑core PCB - and why we think it is quietly becoming one of the most important battlegrounds in India's energy‑transition stack.

Where the Heat Really Hides

A typical high‑power LED still converts only about 30–45% of the electrical input into light; the remaining 55–70% turns into heat concentrated at a microscopic junction on the chip. For every 10°C increase in junction temperature, lumen output usually drops by 1–2%, electrical efficiency slides by another 0.5–1%, and expected lifetime can fall by ~40% once you climb above the rated operating band.

All of that heat has one path out: junction → package → copper traces → dielectric layer → aluminium base → heat sink → air. Most of the stack has been engineered aggressively over the last decade, and yet the weak link is the dielectric. In mainstream metal‑core PCBs (MCPCB), this layer is 100–200 microns thick and delivers only about 0.5–1.5 W/m·K thermal conductivity, yet it sits directly under the hottest point in the system. It is the narrowest section of the pipe.

India's LED Stack: Local From Midstream Down

India has already localised much of the downstream LED stack. Over 90% of consumer LED bulbs are assembled domestically, with Indian EMS and OEMs handling drivers, heat sinks, casings, and final integration. PCB manufacturing now meets around 20–25% of local demand, and copper‑clad laminate production is being built out under recent electronics manufacturing schemes.

The pieces that still largely arrive in containers are the chips and the critical dielectric systems within substrates. LED chips are mostly imported from Taiwan, China, and Malaysia. Dielectrics for MCPCBs show up as pre‑made laminates with supplier‑owned chemistries.

When you zoom out to the system level, the substrate line item is not trivial. A bottom‑up sizing across four LED segments in India (public infrastructure, industrial and large commercial, automotive exterior modules, and residential) puts the MCPCB/IMS market at about ₹1,168 crore per year. The breakdown roughly looks like:

• Public infrastructure (street, high‑mast, tunnel): ~₹73 crore TAM
• Industrial and large commercial lighting (high‑bay, flood lights, warehouse linear): ~₹165 crore TAM
• Automotive exterior modules (headlamps, DRLs, tail clusters): ~₹108 crore TAM
• Residential (bulbs, downlights, panels): ~₹822 crore TAM

A top‑down view cross‑checks this. LED systems across these segments add up to about $4.15 billion in annual revenue, and MCPCB/IMS content typically accounts for around 5% of system value, implying a substrate TAM of about $200 million annually in India alone.

Globally, metal‑clad and MCPCB markets are projected to grow at ~6.2% CAGR into the early 2030s. India's own LED lighting market is expected to keep growing steadily through 2030, adding several billion dollars of value as retrofits continue and industrial and smart‑city upgrades pick up.

Policy, Demand, and Timing

The policy narrative has moved from "Let's get LEDs into homes" to "Let's own the stack that makes them". The PLI scheme for white goods, which explicitly covers LEDs, targets an increase in domestic value addition in components such as drivers, PCBs, LED engines and mechanical parts from roughly 25% to 75% by 2029. Manufacturers that invest in local component capacity can earn 4–6% incentives on incremental sales over a base year.

On the quality side, the Bureau of Energy Efficiency has made QR codes mandatory on LED packaging from March 2025, enabling real‑time verification of star ratings and performance and making it harder for sub‑spec imports to slip through. UJALA has already distributed more than 36 crore LED bulbs, while the Street Lighting National Programme has installed over 1.34 crore streetlights.

As India's carbon‑credit trading framework matures, highly efficient lighting and power‑electronics platforms become levers for monetizable carbon savings, beyond capex and opex gains.

The Barbell Inside the Substrate Market

At one end, global majors such as Henkel, Bergquist Thermal Clad, TTM Technologies, Laird Thermal Systems, and Schweizer Electronic supply high‑performance IMS and thermal PCB systems, often with high‑voltage isolation for automotive and industrial applications. They are priced accordingly and typically reach India as imports for specific, high‑margin programmes.

At the other end, vertically integrated laminate producers such as Kingboard supply low‑cost MCPCBs based on older plastic or epoxy dielectrics. Domestic PCB houses (Ascent Circuits, AT&S India, Shogini Technoarts, Genus Electrotech, PCB Power, among others) assemble aluminium‑backed boards at scale for LED OEMs but rely on imported laminates and do not control dielectric chemistries.

The result is a missing middle. There is very little that is both high‑performance and priced for mainstream wattages in India, and almost nothing that is designed and manufactured domestically at that layer.

What the winners in this layer will look like

The companies that matter here will be materials and process platforms that happen to ship substrates. A few common features are already visible in the data.

They own the chemistry and the factory

The critical edge lies in controlling the dielectric formulation and the printing and curing process that turns it into a thin, reliable layer. In one printed‑dielectric IMS archetype we have evaluated, dielectric thickness falls to about 50–75 microns, thermal conductivity rises to around 2.5–2.9 W/m·K, and junction temperatures drop by up to 20°C in representative LED modules.

They plug into the value chains that already exist

Thermal substrates do not win by building their own brands of streetlights. They win by becoming the default board inside someone else's. In India, that means shipping drop‑in boards into MSME and mid‑market LED manufacturers as well as larger brands like Bajaj, Wipro, Syska, Varroc and Minda, with no changes required to housings or optics.

They compete on performance and cost at the same time

In one costed BOM comparison we have seen, a high‑performance printed IMS board comes in at about ₹560 versus roughly ₹714 for a conventional MCPCB—a cost reduction of around 22% at the board level—while improving thermal performance.

They build a platform, not a single‑use product

LED lighting is an excellent beachhead. Once the substrate and process platform is proven there, the same dielectric stack can be tuned into EV chargers, motor controllers, industrial drives, drones and wide‑band‑gap power modules, where the willingness to pay for thermal headroom is even higher.

This is the kind of market Transition VC likes: a hard engineering problem, a large domestic demand base, policy pull for localisation, and a cost curve that can improve while performance improves. The chip is ready to do more. The dielectric is not. That gap is where the next company gets built.

Inside the Drone Brain

Before a drone can do anything interesting, a quiet conversation must happen between silicon, sensors, and radio waves.

At the core is the flight controller: the board that fuses IMUs, barometers, GNSS and pilot inputs into thousands of control decisions per second. Around it sits electronic speed controllers that modulate motor power, long‑range RF links that carry commands and telemetry, and positioning modules that keep the vehicle where it is supposed to be in three-dimensional space.

In a mission‑critical setting, this avionics stack has a non‑negotiable job: keep the aircraft stable, responsive, and recoverable even when individual components misbehave or the RF environment turns hostile.

ZeroDrag starts from this engineering reality.

Avionics First, Platform Next

ZeroDrag builds the electronics layer as a cohesive system rather than a shelf of parts.

On the RF side, they design long‑range, bi‑directional links that have been exercised over tens of kilometres, with both NDAA‑aligned and commercial variants so integrators can design to different compliance regimes without changing the rest of the stack. On the control side, their autopilot family spans from disposable, single‑use boards to high‑end controllers with dual IMUs, dual power domains and advanced navigation - all sharing a common, solder‑free, connectorized footprint.

The same philosophy shows up in propulsion and navigation. ESCs are built around 32‑bit architectures with proprietary firmware aimed at squeezing more useful flight time out of the same airframe, while GNSS modules combine RTK‑grade positioning with IMU, compass and barometer on a CAN‑bus backbone for long‑distance, noise‑resilient integration.

The result is an avionics kit where mechanical layout, electrical interfaces and software expectations are aligned by design, not negotiated one integration at a time.

The System Problem in Drones

Most drone stacks today are assembled like custom PCs from a decade ago: different vendors for the motherboard, graphics card, storage, power supply, and drivers — except here, the machine is supposed to fly.

OEMs routinely source flight controllers, radios, ESCs, GNSS, payloads and ground systems from unrelated suppliers with different design assumptions and support philosophies. That fragmentation shows up as brittle integration, inconsistent reliability, and long debugging cycles whenever a subsystem changes revision or goes out of stock.

Overlay that with policy. India has explicitly discouraged imports of fully built drones, while still allowing controlled imports of sub‑components to encourage domestic assembly and R&D. In the US, NDAA restrictions and related rules push defence and federal buyers away from certain foreign platforms. Both directions point to the same outcome: demand for trusted, interoperable, non‑Chinese avionics is only going one way.

ZeroDrag's Architecture Bet

ZeroDrag's core bet is that the industry will standardise around a few robust avionics ecosystems rather than perpetually stitching together bespoke stacks.

To earn that position, they have made three important engineering choices:

• Full‑stack coverage. Instead of focusing on a single board, they cover comms, autopilots, ESCs, GNSS and, over time, ground stations, FPV goggles and specialized payloads.
• Solderless, standardized integration. Connectorized layouts and a common 30.5×30.5 mm‑style footprint cut assembly times from hours to well under an hour.
• SWaP‑first design. By aggressively shrinking board size, weight, and power draw, they free up budget for payloads and endurance - especially valuable for ISR and other defence use cases.

On top of that sits proprietary IP: ESC firmware tuned for longer flight times than typical commodity controllers, dual‑band RF designs to better tolerate jamming and interference, and redundancy‑first layouts across the stack.

Why This Fits TVC's Lens

At TVC, we like companies that do two difficult things at once: solve a hard engineering problem and sit in the flow of a structural transition.

On the engineering side, ZeroDrag is doing the unglamorous work of turning fragile hobbyist‑style stacks into something that looks and behaves like avionics. On the market side, policy in India and abroad is pushing buyers to look for secure, non‑Chinese electronics they can standardise on without sacrificing performance or economics.

We also like that ZeroDrag empowers OEMs rather than competing with them. By focusing on subsystems instead of full drones, they position themselves as an enabling layer that integrators can build on.

The Team Behind the Stack

Zero Drag's founding team blends years of hands‑on UAV and avionics work with deep PCB design and systems engineering experience. The CEO brings a mechanical and UAV background and leads product and R&D; the COO comes from a long stint in IT and operations, anchoring procurement, finance and compliance; the CTO has decades in PCB and electronics design, driving miniaturisation and reliability.

We have been tracking the founders since before this round, and have seen them iterate through prototypes, shipping products and a clearer articulation of the "full avionics stack" thesis over several years.

What Has to Go Right

This is not an easy business. Certification regimes, export controls, evolving standards and entrenched incumbents are all real constraints.

But the direction is clear: the world is asking for trusted drone electronics with better integration, better lifecycle guarantees, and less geopolitical risk in the supply chain. We invested because we believe ZeroDrag has a real shot at being that partner - quietly powering missions from inside the drone's brain.

The Market Has Moved Beyond "Can It Fly?"

That shift matters because drones are no longer just toys, experiments, or niche industrial tools. They are becoming part of the infrastructure of defense, agriculture, logistics, surveying, and surveillance. A drone used to spray crops on a farm in Maharashtra has very different requirements from one used to inspect a power line, map a mine site, or support a border-security operation; but in each case, the buyer cares less about the novelty of flight and more about whether the machine works consistently, can be repaired quickly, and doesn't die halfway through a mission.

If a consumer gadget fails, it is annoying. If a drone fails in the middle of a crop-spraying run, a survey mission, or a defense deployment, the cost is operational, financial, and sometimes strategic.

For defense buyers in particular, the question goes one level deeper. They are also asking how resilient the supply chain is, and how much of the platform depends on foreign nations for critical subsystems.

How US and India Policies Are Reshaping the Drone Supply Chain

India and the US have both moved from watching the space to actual policy action. In India, the ban on imports of fully built drones helped create room for domestic manufacturers to emerge, as evident with visible local drone players such as ideaForge and Garuda.

That shift has been reinforced by the ₹120 crore PLI scheme for drones and drone components, which rewards value addition in India. Together, the import restriction and the incentive structure are protecting local assembly and pushing the ecosystem toward deeper domestic capability in the underlying electronics stack.

The US has taken a different route, but the destination is similar. Through NDAA-linked restrictions and the American Security Drone Act, procurement is steadily moving away from drones and subcomponents tied to covered foreign entities, especially Chinese-linked suppliers.

The Real Challenge Is the Stack

For all the progress in drone adoption, the underlying industry is still deeply fragmented. A manufacturer may source flight controllers from one vendor, communication links from another, ESCs from a third, GNSS from a fourth, and payload systems from yet another. Each part may work on its own, but getting them to work together reliably is the real challenge.

Once a drone moves from prototype to production, the buyer is evaluating the platform's ability to stay up, stay supported, and stay replaceable over time. What looks like a parts business is really a systems business, and the systems business is decided by reliability, supportability, and control.

The gap in the market

The market today is split across three familiar categories.

• Hobby-grade vendors are great for prototyping and low-stakes use, but they're not built for harsh environments or long deployment cycles.
• High-end global players have strong products but price points that are hard to roll out at scale, especially for cost-sensitive markets or high-volume OEMs.
• Chinese suppliers historically offered the best mix of cost and capability, but they're now running into policy, procurement, and trust walls.

That leaves the most interesting part of the market still underserved: the customer who wants something affordable, but not fragile; compliant but not overpriced; technically serious, but still practical to buy, integrate, and support.

The deeper gap is in the underlying electronics stack – ones that can provide reliable avionics, communications, controllers, propulsion electronics, and integration layers that are built for scale, can pass customer qualification, and can be supported locally over the life of the platform.

This is exactly the gap that specialized avionics component manufacturers fill. By owning the hard parts — component selection, board revisions, firmware edge cases — they free drone manufacturers to stay focused on what they do best: building great aircrafts.

The Real Moat: Becoming Part of the Build

The companies that are gaining importance in the sector are the ones building the underlying electronics and avionics layer - the companies that make it easier for drone manufacturers and integrators to design, test, manufacture, and scale reliably and cost‑effectively.

The opportunity is to become part of the operating architecture of the ecosystem. Once a company supplies multiple layers of the stack, it becomes embedded. That embeddedness is the moat. It reduces switching, simplifies procurement, and makes long-term support more valuable.

That is the real reset in drones. The market is no longer asking who can make a machine that takes off and lands; it is asking who can build the trusted electronics stack that survives qualification, supports fleet deployment, and holds up under policy, procurement, and operational scrutiny. That is where the category's defensible value will concentrate, and that is where the next generation of drone companies will be built.

In Part 1, we argued that carbon markets are not broken - they are recalibrating around trust. In Part 2, we extended that idea further: markets only scale when they behave like infrastructure, not fragmented transactions.

This is where regenerative agriculture becomes critical. Regenerative agriculture is one of the few nature-based pathways that can scale meaningful carbon removal with established science and visible demand.

In India, this is a systems problem - fragmented landholdings, weak baseline data, and high verification costs make most projects either small or unreliable, resulting in a structural gap: credible supply remains scarce even as demand consolidates around high-integrity, nature-based credits.

We invested in Prithu because they are solving this problem where it exists - on the ground.

Regen Ag's Scale Challenge

Regenerative agriculture holds unique promise for India: 7-9 million hectares already transitioning, with practices like reduced tillage, cover cropping, and agroforestry sequestering 2-4 tCO2e per hectare annually. Yet scaling remains elusive. Over 85% of farmers hold under 2 hectares, creating coordination challenges for monitoring, compliance, and baselines in variable microclimates.

This positions Prithu in a structurally scarce supply segment - credible nature-based credits - amid global demand shifting to 500-700 MtCO2e post-2027, driven by CORSIA and Article 6.

Most projects either cut corners for speed or stay rigorous but small. Prithu rejects that trade-off, prioritizing execution in this unstructured reality.

Field Execution as Foundation

Prithu starts with farmer aggregation via localized field representatives managing 400-500 farmers each. Onboarding takes 10-20 minutes with geo-tagged records, with a fully loaded onboarding cost ~10x below industry benchmarks.

This enables ongoing engagement: practices are observed, reinforced, and adjusted, not assumed post-onboard. Deviations surface early, keeping data dynamic and audit-ready.

Execution is the moat - ~10x lower onboarding cost, 400-500 farmers per agent, and near-zero attrition signal capital efficiency that turns pilots into scale. This density de-risks issuance (90-95% odds post-registration) and creates compounding year-on-year cashflows per hectare.

GTM Built for Farmers

Prithu's go-to-market centers farmers: 40% revenue share via direct benefit transfers, performance incentives for reps, and ties to FPOs for trust. Near-zero attrition reflects clear economics and sustained support.

It's vertical control from origination to sale, across 43,000 hectares onboarded - this density turns fragmented land into scalable projects.

Farmer-first GTM aligns incentives and leverages district-level access for rapid ramp - 43k ha live, 5 lakh ha coming up.

MRV in Service of Execution

Prithu's digital MRV supports this: geo-tagged field proofs, satellite validation, and hashed trails compress timelines and boost issuance odds to 90-95% post-registration. It's hybrid by design - tech aids redundancy, but field oversight remains core.

Once registered (12-15 months), projects yield credits for 25-30 years at falling costs, creating long-duration assets.

Durable Expansion, Regen Core

Regen-ag drives volume; biochar partnerships add permanence, capital-light, leveraging existing networks for soil application and verification.

Diversification balances volume with upside of cashflow, mitigating risks, and preserves ROIC as the platform scales to its cumulative potential.

Carbon markets scale when reliable enough to underwrite here, in India's complexity. Prithu builds that system through farmer focus and execution, proving that if it works amid fragmentation, it works anywhere.

India's chaos is climate tech's proving ground. Prithu is building credible supply here, redefining markets - one farmer cluster at a time.

In Part 1, we argued that carbon markets have a trust problem, not a demand problem. Buyers are still here. What they are no longer willing to do is buy credits whose integrity depends on fragile baselines, indirect measurement, and assumptions that cannot withstand scrutiny.

Part 2 asks the harder question: what does it take to build a market that can scale, and deserves capital.

What A Functional Carbon Market Requires

If measurement improves, baselines standardise, and risk is priced explicitly, carbon credits stop behaving like fragile instruments and start behaving like infrastructure.

That shift is not semantic. It changes who participates, how capital flows, and what scale becomes possible. Today, transactions are episodic - short-term, inconsistent, and fragmented. The market clears, but it doesn't compound.

First, credits become underwriteable

When measurement is continuous, and uncertainty is explicitly modelled, credits can be evaluated like any other risk-bearing asset.

• Cash flows can be forecast with defensible assumptions
• Reversal risks can be priced into contracts
• Long-term performance can be monitored, not assumed

Second, contract structures evolve

Instead of spot purchases or short-term offtakes, the market moves toward:

• Long-duration offtake agreements (10–20 years)
• Forward contracts linked to future credit generation
• Blended finance structures with layered risk tranches

Third, capital costs compress

Today, carbon projects often carry high implicit risk premiums because outcomes are uncertain and verification is intermittent. As systems mature, better data reduces perceived risk, standardization improves comparability, and liquidity deepens secondary markets.

Fourth, the market deepens beyond compliance and CSR budgets

Once credits are credible financial instruments, they move out of sustainability silos and into core capital allocation decisions - infrastructure funds, insurance capital, sovereign and pension capital.

India: The Real Stress Test

India is where that question gets real. Fragmented smallholders, policy overlays, uneven data, and the need to aggregate execution at scale make it the hardest place to build this market and the most revealing one.

Building Carbon Markets in India

A functional carbon market in India will not look like a simplified Western model scaled up. It will be built around constraints and will derive its strength from handling them.

1. Aggregation at Scale, Without Losing Fidelity

India's carbon market will be built on smallholders, using FPOs and cooperatives as primary aggregation layers, standardized digital interfaces for farmer data capture, and geo-tagged activity records integrated into MRV systems.

2. MRV Built on Digital Public Infrastructure

India already has the rails: satellite data (ISRO) for land-use monitoring, weather and soil datasets for baseline calibration, and Aadhaar-linked systems for identity and payments.

3. Making Farmer Economics Work

Direct Benefit Transfers linked to verified outcomes, clear revenue-sharing models, and predictable payout timelines tied to credit issuance cycles.

4. Designing for Policy, Not Around It

Additionality frameworks that explicitly account for existing schemes, baselines periodically recalibrated to reflect policy-driven adoption, and projects structured with buffers against regulatory shocks.

5. Standardization Without Oversimplification

Success lies in bounded standardization: region-specific baseline libraries, methodologies tailored to agro-climatic zones, and shared data layers with localized calibration.

6. A Tiered Market Will Emerge

A mature Indian market will likely stratify into high-integrity credits, mid-tier credits with moderate verification intensity, and commodity credits with lower assurance and pricing.

7. Built for Global Capital

India's carbon market needs to plug into global demand while maintaining local integrity - alignment with international standards, acceptance by global buyers, and export-grade credits that command premium pricing.

What Success Actually Looks Like

Success is not the elimination of uncertainty; it's the honest handling of it. A working system measures uncertainty and prices it transparently, replaces periodic validation with continuous observation, aligns incentives across developers, verifiers and buyers, and integrates policy as a core variable.

India is the most important proving ground because it compresses the sector's hardest questions into one operating environment. That is the real test: can the carbon credits market be structured well enough to deserve scale.

We all have trust issues with carbon credits. Fair enough. Most of them are earned.

But the popular narrative that demand is fading and that corporates are quietly backing away is lazy analysis. Demand isn't disappearing, it's evolving.

What's actually happening is more structural: buyers are still here, but they've stopped trusting what they're buying.

The Myth of Fading Demand

Everyone loves a clean collapse story. Carbon markets make for an easy target: inflated claims, shaky projects, and a few high-profile blowups.

Corporate net-zero commitments are still climbing. Retirements haven't fallen off a cliff; what has changed is buyer behaviour. MSCI's 2025 State of Integrity report lays it bare: corporate net-zero pledges surged 227% while retirements dipped just 7%. Fastmarkets clocked 2025 retirements steady at 169 million tCO₂e.

But here's the arithmetic that matters: global net-zero demands ~6 GtCO₂e annual removals by 2050. Renewables and electrification cover ~30-50% of the needed reductions. The rest cannot be fully electrified or renewed away. Without credible carbon credits, the Net Zero 2050 targets don't add up.

Buyers understand this existential math. Procurement teams now diligence credits like derivatives trades. The result is a barbell market: AAA-rated credits clear at 4x premiums, everything else struggles for bids.

The Mechanics (and the Illusion of Precision) of Carbon Credits

At a basic level, a carbon credit represents one tonne of CO₂ equivalent - either avoided or removed.

• Avoidance credits: solar displacing coal, cleaner cookstoves reducing biomass use
• Removal credits: soil carbon, forests, direct air capture

The system assumes you can measure, verify, and issue a tradable unit. This stack has a name: MRV - Measurement, Reporting, and Verification. The problem is step one.

Carbon is not directly observable. It's inferred using proxies such as satellite imagery, soil samples, activity data, and baseline assumptions. Each layer introduces uncertainty.

Industrial projects operate within tighter bounds - you can get within a 10–20% error band. Nature-based credits are the real plot twist.

Nature-Based Credits: High Potential, High Variance

This is where the system starts to break down.

Take regenerative agriculture in India. Farmers adopt no-till practices, cover cropping, and organic inputs. Soil carbon increases over time. Credits are issued based on the incremental gain over a baseline.

Soil carbon varies by rainfall patterns, crop cycles, microclimates, and soil composition. Measuring it requires multiple soil samples per hectare, longitudinal tracking across seasons, and statistical modeling across fragmented plots.

Now layer in India's structure: 85% smallholder farmers (<2 hectares), highly fragmented landholdings, policy overlays, and monsoon-driven variability. A clean dataset doesn't exist. What exists is a probabilistic system.

So, What's Actually Broken?

Strip away the noise, and the failure modes are clear.

Measurement is indirect, but treated as definitive

We are not directly measuring carbon; we are estimating it through a series of proxies and assumptions. This creates a core mismatch: the system produces estimates, but the market consumes them as facts.

Baselines can be negotiable

Every credit relies on a counterfactual: what emissions would have been in the absence of the project. Small changes in assumptions can materially change credit volumes. When baselines become flexible, supply becomes elastic in the wrong direction.

Incentives are misaligned

Developers optimize for maximum issuance, registries optimize for market activity, auditors optimize for repeat engagements, and buyers optimize for reputational defensibility. None of these incentives naturally converge toward accuracy or integrity.

Verification costs do not scale cleanly

High-integrity credits are data-intensive by design, requiring frequent sampling, cross-verification, and long-term monitoring. This creates a cost structure that does not scale easily in fragmented systems like smallholder agriculture.

Policy risk is ignored (until it isn't)

Subsidies can eliminate the economic need for certain interventions, state programs can drive adoption independent of carbon incentives, and regulatory changes can alter project viability overnight. Most carbon models treat policy as static or exogenous - it is neither.

Taken together, these are systemic constraints, not isolated failures. Carbon markets are the only financing mechanism capable of mobilizing the removals a net-zero target otherwise collapses without. That is why the market's real challenge is not demand, but design.

Part 2: What it takes to turn carbon markets into climate infrastructure. The next frontier is not demand, it's discipline. More Soon.

The Indian power grid is the largest machine most of us will never see. It runs through forests and farmland, deserts and dense cities. It has been stitched together over decades of policy, engineering and political will. And over the next decade, it will either become the backbone of our energy transition - or the bottleneck that holds it back.

This is the system we are choosing to underwrite.

The Evolution of India's Grid

India today runs a single, synchronised national grid. Power that starts life as steam in a pithead coal plant or photons on a Rajasthan solar farm can be dispatched in real time to a data centre in Hyderabad or a pump set in a village in Bihar.

Generation capacity has grown in parallel, from roughly 305 GW to over 500 GW, with significant contributors being solar, wind, hydro and nuclear – in fact, in 2025–26 alone, India added over 50 GW of new capacity, around 80 percent of it from renewables.

The Systemic Vulnerability of Distribution

While the transmission and generation sectors show clear progress, the distribution layer remains the most fragile link in the chain. State Distribution Companies (DISCOMs), despite supplying nearly 90% of India's electricity, remain in deep financial distress.

Between 2015–16 and 2022–23, their accumulated losses rose by about 81% and debt by around 62%, reaching roughly ₹6.8 lakh crore and ₹6.6 lakh crore respectively.

Aggregate technical and commercial (AT&C) losses have fallen at the national level from about 25.5% in FY13 to around 15.4% in FY25, but they remain far above global benchmarks and exceed 30–40% in some states.

The Transition to a Data-Intensive System

For a hundred years, the grid was an electromechanical system: meters turned dials, linemen read them once a month, and then DISCOMs turned the readings into bills.

Instead of one reading per meter per month, discoms now collect 96 interval readings per day—15‑minute data streams that capture not just how much power a consumer used, but when, at what voltage, with what power quality. Scaled to hundreds of millions of meters, this is a 3,000‑fold increase in data volume that pushes large utilities into the petabyte era.

This means software—not just steel and copper—is becoming grid infrastructure. This final layer is where the energy transition moves from a "hardware story" to a "high-margin software story."

The Emergence of the Orchestration Layer

Historically, utility software has been built in rigid, vertical layers: the Head-End System (HES) at the edge, Meter Data Management (MDM) in the middle, and customer information/billing platforms at the end. Each layer has its own database, schema, and vendor, and even basic operational questions require stitching together multiple systems.

Autonomy through the Agentic Layer

Once this layer is in place, the system moves from being observed to being managed. AI agents compare energy flows in real time and identify discrepancies. Predictive asset management monitors patterns of degradation and recommends interventions before breakdowns occur, potentially extending transformer life by 2–3 years. Operators can define outcomes at a high level rather than executing tasks manually.

Why is this the Investable Wedge

At this point in the stack, the source of value shifts from building infrastructure to operating it. The constraint is no longer generation or capacity; it's execution.

Even small improvements in loss reduction or collections translate into thousands of crores in recovered revenue at the system level. These gains determine whether DISCOMs can pay generators, service debt, and reinvest in the grid.

Enter: WorkOnGrid

WorkOnGrid operates as a unifying software layer across the utility stack. It ingests data from fragmented systems such as HES, MDM, billing and ERP, normalises it into a single data layer, and creates a continuous feedback loop between what is happening on the network and what actions need to be taken.

In live deployments, this approach has demonstrated the ability to isolate high-impact loss pockets, improve collection efficiency by 3–5%, and reduce AT&C losses by up to 2% points within months. At system scale, these are the difference between a solvent and an insolvent grid.

WorkOnGrid does not compete with the existing stack; it makes it usable. And as it becomes embedded across data, workflows, and operations, it moves from being a software product to becoming core infrastructure within the utility.

Today, we're talking about carbon capture, more specifically, about how carbon capture evolves into industrial manufacturing.

The central question behind this investment was simple: Can carbon capture integrate into heavy industry in a way that strengthens business fundamentals rather than weakening them?

The Structural Gap in Carbon Capture

For the last several decades, the dominant carbon capture technology has been the amine system. An amine system uses chemical solvents that absorb CO₂ from flue gas streams, chemically binding to carbon dioxide before being heated to release the CO₂ in concentrated form.

It works. The chemistry is well understood. So why hasn't adoption scaled? Because the economics create a structural gap. Installing an amine system means spending large upfront capital ($60–100 per tonne), consuming 20–30% of plant output to run the system, and paying to compress, transport and store CO₂ without generating new product revenue.

In India, where cement and steel operate on thin margins and where there is no widespread CO₂ storage infrastructure, the financial incentive is weak. Carbon capture reduces emissions, but it does not improve the income statement.

The Biological Lever in CCUS

The biological lever introduces a different idea: instead of separating CO₂ and burying it, convert it into something valuable. Microalgae naturally fix carbon through photosynthesis, converting roughly 1.8–2x their weight in CO₂ into biomass containing proteins, pigments, lipids, and specialty compounds with established markets.

Previous biological approaches faced real-world constraints: industrial flue gas fluctuates between 8–15% CO₂ and contains impurities. Many algal strains struggled above 35°C, while Indian industrial zones regularly reach 48–50°C. Productivity often stayed at 0.2–0.3 g/L/day. The biological pathway was promising, but the architecture was not built for industrial volatility.

Enter Intrinsic Foundries

What Intrinsic Foundries did differently was design the system around industrial conditions.

They developed a carbon banking mechanism that buffers CO₂ as bicarbonate within the growth medium, storing up to 6 g/L equivalent per liter. They curated a non-GMO strain library tolerant to high CO₂ concentrations, impurities, and ambient heat up to 50°C. They also redesigned the reactor manufacturing through on-site PMMA extrusion, reducing per-kL capex by roughly 60% at scale.

The shift is subtle but important: the system adapts to the factory, not the other way around.

Why This Matters to Transition VC

Here are Gaurav Patil's thoughts:

At Transition VC, we invest in engineering-led infrastructure that can stand on its own economics. Our mandate has always been clear: back companies that build real systems for the energy transition, particularly in markets where capital discipline and operating constraints are non-negotiable.

We focus on platforms that do not rely solely on subsidies, sentiment, or regulatory timing. We look for architectures where environmental impact compounds because the business model compounds.

Intrinsic Foundries fits squarely within that lens. India, and the broader Global South, will not decarbonize through imported solutions optimized for different cost structures. The transition here will be built by companies that understand heat, volatility, capital efficiency, and margin pressure.

To us, Intrinsic doesn't simply promise carbon utilization, they have the clarity of integration: capture embedded at point source, modular hardware designed for industrial environments, and downstream value creation that strengthens the income statement of the emitter. This is the type of platform we believe defines the next decade of industrial decarbonization: Engineering-first, economically aligned, and globally relevant.

And that is precisely why we are backing them.

In our earlier piece, The Cooling Bottleneck Inside Industrial India, we argued that industrial cooling in hot, humid climates is not a comfort problem but an infrastructure bottleneck. Cooling systems are being asked to do two jobs at once: pull down temperature and strip out moisture; and the way we do that today is expensive, energy-intensive, and poorly suited to the next wave of Indian manufacturing.

Albatross Energetics was the first company we found that solved this problem from first principles. This is why we invested.

From the bottleneck to the bet

We made three claims that shaped how we looked at this space:

First, most Indian factories operate in hot, humid climates where conventional HVAC systems become increasingly inefficient. Today's choices are all compromises: use solid desiccant wheels and accept high regeneration energy, rely on refrigeration and sacrifice humidity control while increasing electricity consumption, or deploy imported liquid-desiccant systems that are expensive and poorly suited to Indian operating conditions.

Second, humidity control is becoming a production requirement rather than a facility preference. Pharmaceuticals, food processing, textiles, solar manufacturing, semiconductors and advanced electronics increasingly depend on tightly controlled moisture levels.

Third, reducing temperature through the typical refrigeration cycle and reducing humidity separately through desiccants is proven to be significantly more energy efficient than doing both together through the typical refrigeration cycle.

Albatross sits where those three lines intersect: it builds industrial cooling and dehumidification systems around a proprietary liquid desiccant - a non‑toxic, non‑corrosive chemistry that decouples humidity removal from temperature control and runs more efficiently than both solid desiccants and conventional vapour‑compression systems in humid conditions. More importantly, it solves the problem that has prevented liquid desiccants from becoming mainstream for decades: chemistry.

Why decoupling cooling and humidity is the real unlock

Air conditioning has two jobs: making air colder, and making it drier. In a standard HVAC system, one refrigeration loop does both. That's fine when the air is only moderately humid. At 70–80% relative humidity, it gets expensive — the compressor is now working overtime to wring water out of the air, not just to move heat.

To remove moisture, the refrigeration system cools air below its dew point so water condenses on the cooling coil. The air is then colder than required, so it must be reheated before entering the conditioned space. In other words, the system spends energy cooling the air further than necessary, only to heat it back up again.

Desiccant systems flip the order: dry the air first, then cool it. A desiccant material absorbs the moisture directly, which leaves the cooling loop with only one job — lowering the temperature. No overshooting, no reheating.

What Albatross actually builds

Every Albatross product is built around two things: a custom heat-and-mass exchanger (HME) and a proprietary liquid desiccant.

Step 1 — Drying and cooling the air

Humid air enters the machine and flows over surfaces coated with a thin film of liquid desiccant. Chilled water runs inside those same surfaces. The desiccant pulls moisture out of the air; the chilled water pulls out the heat. The air that comes out the other side is both drier and cooler — in one pass.

Step 2 — Recycling the desiccant

The desiccant is now full of water, so it moves to a separate section where hot water (roughly 45–65°C) gently heats it. That heat pushes the water back out into an exhaust air stream, and the dried desiccant returns to the main loop to be used again.

The key point is that 45–65°C is a low temperature. So the regeneration step can run on heat pumps, solar thermal, or waste heat that a factory already has — no high-pressure steam required.

The two product lines today

• Industrial liquid-desiccant AHUs - turnkey air-handling units for factories and clean rooms that need tight control over humidity and temperature.
• OEM integration kits - absorber and regenerator modules plus the chemistry, sold to AHU and DOAS manufacturers who want to add liquid-desiccant capability to their own equipment.

Together, these models allow Albatross to sell both complete systems and the core technology platform, expanding its addressable market beyond direct equipment sales.

Solving the toxic, corrosive liquid‑desiccant problem

Liquid desiccants are not a new idea; they have been explored for decades because, on paper, they are very effective at handling humidity. The reason they have remained niche is that the usual suspects - lithium chloride, lithium bromide - are both corrosive and, in many cases, hazardous.

Toxicity shrinks the addressable market because you cannot put an aggressive chemical loop anywhere near people or sensitive products. Corrosion shortens equipment life and forces titanium and speciality plastics into what should be standard steel‑and‑aluminium hardware.

Albatross's chemistry is different. The company has developed a blend of organic solutions that maintains strong moisture absorption performance, regenerates at low temperatures, and behaves far more benignly toward common metals. Independent toxicology work suggests low acute toxicity and no major chronic hazard classifications; corrosion tests indicate acceptable performance on aluminium over long periods.

Those are not cosmetic tweaks; they change where and how you can deploy liquid desiccants:

• The same loop can now run in environments like pharma, semiconductor, solar cell, textile, and food environments where people and products are present.
• The absorber and regenerator (main hardware) can be built largely from aluminium and standard alloys, which are cheaper, lighter, and straightforward to fabricate and service in India.

In other words, Albatross has solved for the chemistry, materials, and manufacturability as one problem, rather than separate checkboxes.

Our previous article mapped the space into three incumbent technologies: Refrigeration‑based HVAC, solid desiccant systems, and traditional liquid desiccants. Albatross effectively creates a new corner of that map.

• Vapour‑compression HVAC (the standard chiller and coil setup) is cheap and widely available. It is excellent at straightforward cooling but inefficient for deep drying because it has to over‑cool and reheat.
• Solid desiccant wheels achieve very low humidity levels and are entrenched in pharma and specialty manufacturing, but they require high-temperature (120-150°C) and add extra heat that must then be removed again.
• Conventional liquid desiccants promise thermodynamic advantages and low‑grade heat compatibility, but toxicity and corrosion fence them off to narrow, high‑spec niches.

Albatross's system offers deep drying at lower regeneration temperatures, with non‑toxic, non‑corrosive chemistry and the ability to drop into the existing cooling stack rather than replacing the entire system. That combination is what turns liquid desiccants into a general-purpose building block for industrial and large commercial dehumidification.

We invested because Albatross combines three things that rarely exist together: a meaningful thermodynamic advantage, chemistry designed for commercial deployment, and a system architecture that integrates into existing HVAC infrastructure rather than replacing it. We think that combination positions liquid-desiccant cooling to move beyond niche applications and become a foundational layer of industrial climate control.