It is the same question flipped two different ways. The question arises because businesses seek profitability and product excellence, creating barriers to entry in an industry that is fast becoming commoditized.
Large manufacturers in the ICE business outsource everything but keep tight control over the quality and final assembly of the products. Break down any mass-produced motorcycle, and almost every significant component would be outsourced. Yet, they are not off-the-shelf and unavailable in the open market for others to buy.
Outsourced vs Off-the-shelf (OTS)
Outsourced does not mean OTS. In the ICE industry, there is hardly anything OTS. In contrast, in the E2W industry, there is a possibility of a lot of OTS, especially if you aspire to be a niche manufacturer of handcrafted motorcycles in a European village, taking pre-orders on Indiegogo. That’s not scoffing at anyone – the world is a more beautiful place because of niche manufacturers, and Niu started life on Indiegogo.
Nearly every component in an ICE motorcycle is designed by the OEM. However, nearly every component is supplied by a specialised supplier. There may be components that seem OTS—tyres, wheels, brake systems, suspension, and ABS systems. Even when the components seem OTS, the OEM integrates them, tests them, and tunes them before they start production. Different models may have similar components but with different mounting points. The 41 mm forks on two motorcycles from the same supplier are likely not interchangeable.
ICE Industry Uses Specialised Suppliers
A large manufacturer like Honda would have a motorcycle frame supplier, which in some cases may be integrated within the Honda assembly plant. The frame suppliers are specialised and generally 5-6 of them cater to the entire industry in a large market like India or Indonesia.
The case with other components is similar. Most large OEMs assemble their engines, but nearly every component comes from a specialised supplier. The same goes for the transmission, where someone supplies the gears, sprockets, and shafts, but the OEM does the final assembly. There are specialised suppliers who cater to a specific family of components.
Similarly, headlamps are never in-house because they require an excellence in optics that suppliers who specialise in lamps have developed. For an OEM to do that is not impossible, but it is challenging and not the best deployment of capital.
The same goes for the cables, wiring harnesses, brakes, ABS systems, wheels, and tires. They have specialised manufacturers supplying to the entire industry. The brakes, the suspension, and many other components are outsourced because there are dedicated suppliers who can do the job much better.
The Difference in the Electric Two-wheeler Industry
Some non-core components—the frame, saddle, brakes, wheels, suspension, and tires—remain fairly standardised and similar to what we see on the ICE side. However, the core components, also the ones with the highest value—the motor, battery, and electronics- are distinctive, and we will look at them in detail.
Before we go further, here is a look at what Ather in-houses vs. what they outsource. Mind you, this is a manufacturer that has developed a rather competent scooter with high-end components all by itself.
Apart from the portable charger, Ather has designed everything in-house. At the same time, apart from the battery pack and the final assembly of the scooter, Ather manufactures nothing in-house. This is indicative of how the in-housing character changes with core components in E2Ws.
At the core, the E2W/L-category EV industry is driven by supplier innovation, not OEMs’ chutzpah.
Cells and Packs
You need cell suppliers to innovate on chemistry, motor suppliers to do the heavy lifting with rare-earth-free motors, and so on. As of now, the globally accepted trend for large manufacturers is to develop the packs, manufacture the packs in-house, and source the cells prudently.
At the very core are the cells, the unit component of the battery. Cells may appear deceptively simple, but are challenging to develop and costly to manufacture. Worse, only very large-scale manufacturing is profitable—Advertising is a reason why cell plants are called gigafactories, not mega-factories or just factories.
Does a Gigafactory Make Sense Today? The Ola Electric Case Study
Last financial year (ended March 2025), the entire Indian E2W industry sold 1.123 million units, or roughly 93,583 units every month. Some months’ sales have spiked beyond 110k levels, but they have soon been followed by terribly slow months.
We estimate that the average battery capacity of each scooter is 3.5 kWh. Let us be generous and assume that for the next two years, the Indian E2W industry will grow at 20% per year. This puts the average annual cell capacity utilization for FY 27 at 5.5 GWh.
Mind you, that’s for the entire industry. In the best case scenario, Ola has a 20% market share potential for the future. That puts Ola’s requirements at 1.1 GWh.
We also must consider that Ola plans to make 4680 cells with NMC chemistry, at least initially. This is debatable when the world is moving to LFP.
Also note that 4680 cells are different in physical dimensions, Voltage, Current, and Energy Density from 2170 cells that Ola currently uses. If all Ola products, present and future, have to move to 4680 cells, the company would need to re-homologate all batteries afresh and, in some cases, also re-homologate the vehicles. The batteries would also change in paper specs. But all that should make sense because Ola would be in-housing the cells, and theoretically, the cell supplier’s 20- 25% margins should come in-house. That makes a lot of economic sense.
But does it?
There are worrying gaps in Ola’s announced plans. A Gigafactory makes money when it crosses a certain threshold. It’s all about volumes, and the minimum needed to make sense for investment is about 20 GWh. That’s what Ola plans to reach eventually. That’s what they have signed up for with the Indian government under the PLI ACC scheme. Considering, in Ola’s best-case scenario, they don’t need more than 1.1 GWh per year, what they would do with the spare capacity remains to be seen.
But in the big picture, the single biggest negative for Ola is what is today the single biggest positive for everyone else – crashing battery prices. Ola is planning to start producing a commodity with huge CAPEX investment, which is crashing in prices globally. There is a likelihood that Ola’s rivals may get cells cheaper from China than Ola would be able to manufacture in-house.
Ola’s biggest cost differential driver for cells was the generous PLI ACC subsidy that the government was offering. However, with deadlines not met, it is unlikely that the company would get any subsidies.
The global prices are crashing because the Chinese have an oversupply, a story repeated before with solar panels and Compact Discs, where the India-manufactured prices could not undercut Chinese FOB prices.
Would they, this time?
While the cells are complicated, the packs can be as simple or challenging as you make them. Here, we should separate the development of a pack from the mass manufacturing of one. Developing a pack is moderately challenging, and a startup in a developing country can have as much of a go at a pack as one in the developed world. However, mass-manufacturing packs need large investments, as elements such as sorting and quality testing cells come into play. The complexity would also be governed by the cell connection process one uses – wire bonding or laser welding. In any case, the mass production of batteries at the rate of thousands every month needs significant investment. Developing the same batteries and even testing them does not.
With E2Ws, there is the possibility of a larger share of OTS components. The cells are OTS for the vehicle manufacturer, governed by quality, steady supply, voltage uniformity, energy density, with/without China supply chain considerations, and a few other factors. In the industry, the cell is very rarely a talking point, so it is okay for a premium manufacturer like Ather to use both Chinese and Korean suppliers.
From the Ather RHP:
We sourced our lithium-ion cells, one of the critical components in our E2Ws, from two foreign suppliers located in China and South Korea…
Ather is a premium mass manufacturing brand in India, accounting for about 12k sales/month and a 12-15% market share. They are fairly large. A customer buying their scooters does not know who the cell supplier is. What they do know is that the batteries are IPX7 rated, and they get a certain warranty on the pack. The cell supplier has no positive impact on Ather’s brand, and over time, the Bangalore-based company would have moved some of its supply chain to China-based suppliers in the hunt for better costs.
The only anomaly to this is when a startup deploys BYD batteries. Some other OEMs also harp on the fact that they use CATL cells, or Molicel, or Enevate, especially in cases where the batteries are fast charging, and the vehicle is a performance motorcycle. Here, the cell brand stands for longevity, fast charging, high energy density, innovation, etc. It makes sense for the OEM to get a positive rub from the brand. Advertising cell suppliers also work for Chinese scooter manufacturers, trying to shun the typical low-quality image as they target global markets.
Should all packs be developed/manufactured in-house?
We have already established that making cells in-house makes zero sense. It is a high-capex item, and small cell factories make little sense. Globally, India-based Ola Electric is the only pure-play two-wheeler manufacturer to have committed to cell manufacturing. We say pure-play because FinDreams, BYD’s cell arm, has likely invested in Spain-based scooter startup Nerva.
However, making battery packs in-house makes sense, depending on the type of pack. Decision making should be driven by the type of pack because while in-housing fixed packs make sense, we find it questionable when every startup starts putting significant capital into developing portable packs. There are multiple portable pack types in every significant market. None of them has any significant advantage over the others.
Would the landscape have been different if startups agreed on universal pack designs and focused on product differentiation in the rest of the vehicle?
Motor
The motor in a commuter scooter is not complex. It revs at mid-four-digit rpms and can have an isolated efficiency of 85-95% for most mid-drive motors. There is very little reason to develop it yourself. Tens of competent suppliers are in every geography, all willing to pass you scale efficiency if you are worthy. At the same time, several reasons may push an OEM to develop and manufacture its own motors.
- Develop motors yourself if you are attempting one with fewer or no rare earths. That is a game-changer and a worthy effort. It reduces BoM costs and the dependence on a China-based supply chain.
- You are developing a performance machine with a high-voltage architecture. You need a high-revving motor, say 15,000+ rpm, and there are not too many of them around OTS. Developing your own makes sense.
- You want to integrate a gearbox with the motor. Not too many OTS options come with that.
- There are physical attributes – diameter or length – that you can innovate on to deliver higher power or smaller size than the OTS options available.
- The same goes for integrating motors with gearboxes or integrating motors with controllers, within the same casing. This is a BoM cost-saving measure as it saves material costs by reducing a few centimetres of HV cable.
- The suppliers don’t have capacity to ramp up quickly, or you feel they are not optimizing the price. Both are highly unlikely scenarios.
- If you can control the motor, especially the motor controller part of the equation, updating software, fine-tuning things, improving throttle response etc. becomes easier and faster as its your internal team working on that. Many weeks of operation and many links in the management chain are removed.
- You have a product plan that needs a wide range of motors, and you want to optimise on every motor’s input raw material costs and not optimise them through software alone.
- You can deliver a better torque curve than OTS motors. However, this is more an attribute of the controller than the motor, and we will discuss it in the next section.
Motor manufacturing lines are nowhere in the same discourse as cell manufacturing plants. Motor manufacturing is cheap: cell manufacturing needs deep pockets.
Controllers & Software
There are plenty of them around in the vehicle. The most important are the Battery Management System (BMS) and the Motor Controller (MCU). There are plenty of OTS options around but this is one area where startups like to tinker around, and for good reason. Every controller has software that does the ‘controlling’ and optimising the hardware is key to delivering a good product.
The BMS optimises the battery for range, the most critical factor for any electric two-wheeler. At the same time, it ensures that cells discharge and charge optimally, preventing thermal issues.
However, the MCU is more interesting. At the basic level, it can vary the speed, power, and torque output. It can also change the polarity of the motor to allow reverse function. It is also responsible for maintaining the thermal characteristics of the motor and shutting down or reducing power (derating) when needed.
Designing controllers is easy; testing and fine-tuning them is difficult. Manufacturing them at an industrial scale with high quality requires significant investment and should be avoided, as numerous manufacturers worldwide are willing to do this at competitive prices.
But fiddling and tuning controllers is important. It’s the equivalent of learning to walk in the E2W industry. One cannot skip it before moving to bigger things. Designing controllers and fiddling with them teaches the engineering team to speed up the overall product development cycle.
In short, a startup can add flavour to any OTS motor by fiddling with the controller and its software. Manufacturers can completely skip the motor part and focus on the controller.
The critical thing here is the understanding with the supplier. Many suppliers, especially the big ones, don’t like the idea of getting kicked out of the controller part of the business.
The E2W Industry’s Infatuation with Vertical Integration
We have a casual, almost bordering on cavalier opinion that vertical integration comes from every start-up founder blindly aping Musk and Tesla‘s business model. Tesla has an immeasurable but very high vertical integration level, and arguably, most of the company’s profitability comes from that.
A deep dive into Tesla indicates that they design a lot in-house. This includes the software on the cars, some of the chips, the packs, the motors, the cars themselves, and even the factories that manufacture them. Tesla owns and operates all its chargers, experience centers, and sales and service operations.
Apart from designing, Tesla also manufactures a lot in-house. Not everything, but very close to everything.
Tesla drove vertical integration because they have been on a fast innovation train. The motors and other components that they needed, and eventually designed and manufactured, did not exist.
The 4680 cell was Tesla’s innovation. It did not exist already. Tesla needed to push the technology envelope, which compelled them to manufacture everything.
The only exception is seats. They are one of the heaviest components in car assembly, and scaling up seat manufacturing takes time. Seats, especially the motorised ones found on all Tesla cars, are also the most expensive component of a car after the battery and drivetrain. It makes sense to produce them in-house and save the margin that seat suppliers make.
But Scooters are not Cars
Tesla’s move to design and produce in-house was driven by its position as the innovation leader in a high-cost, high-BoM-cost, high-CAPEX industry saturated with old-time players.
It was an industry begging to be disrupted.
When Xiaomi unveiled the SU7 in December 2023, it was hard to miss that the car and the tech Xiaomi promised were all targeted at Tesla. From high(er) revving motors, to CTB, to an integrated die-cast rear floor section (giga casting), to smart interiors, and many more, Xiaomi claimed to take innovation beyond Tesla with the first model only. The first cars may be flawed, but the aspirations aren’t.
And remember, Tesla may have a lot in-house, but it is still not vertically integrated. Not at least at the level of BYD, which has interests in mining, makes everything that Tesla does and much more, makes its own semiconductors, even designs and commissions its own car-carrying ships, and more.
However, a Tesla or BYD in-housing everything makes sense for the massive scale they operate at, the high-cost/high-price products, and the edge they need to maintain. No two-wheeler manufacturer, including Honda or Yadea, has the global scale in E2Ws to justify the vertical integration of the level of Tesla or BYD.
Actually, the entire E2W industry does not have the combined value scale to match even Tesla or BYD in isolation. In 2024, Tesla sold 1.79 million cars, for automotive revenue of USD 77.07 bn, an estimated average transaction price of USD 43,000 per car.
In comparison, an average electric scooter sells at about USD 1200-1500. We estimate global sales of about 9.0 million units. That’s only a USD 12.15 bn market, which is only 16% the size of Tesla.
Mind you, this market is distributed between at least 150 significant brands, Yadea being the biggest one. Even they were worth only about four million units in 2024. At that scale, there is little they can vertically integrate.
Lack of Scale Limits Integration
In many cases, the component you may be thinking of localising in housing would not be easy to develop and design from scratch. Take, for example, microprocessors. Their design needs a completely different skill set and high-quality teams, something that even the biggest two-wheeler manufacturer would find challenging to create as a sub-domain of their overall engineering. Microprocessor designers are some of the most alpha engineers today, and attracting them, in significant numbers, to an organisation whose core is not microprocessor design or computing of any sort is nearly impossible.
Then there is the level of complexity—even the most highly conceptualised two-wheelers with cameras and sensors front and rear would not need chips as complex as those in a five-metre-long electric luxury sedan.
Then there is the BoM Cost Limitation
There is also the BoM limitation. A USD 43,000 Tesla should have a BoM cost of around USD 25,000 (we may be off a few thousand dollars here, but it does not impact the analysis). With 25k dollars, there is a scope to save a few dollars everywhere if you rationalise and vertically integrate everywhere. So, instead of Qualcomm or TSMC designing and manufacturing chips, if you bring the entire function in-house, there is a lot of meat on the bone to chew on, making it sense to set up high-quality teams to run the function.
In comparison, a USD 1200 scooter would not have the BoM cost scope for a high-spec chip, whatever way you may want to slice and dice.
The same goes for every other component. The largest scooter displays and control units would pale in comparison to those on any self-respecting electric car: the cost and the potential to save also diminish accordingly.
Car seats take up a lot of space and incur high logistics costs. A 40-ft container would at most pack 55-60 sets of five seats each. The number drops drastically for seven or eight-seat sets. When you are manufacturing hundreds of cars every day, it makes sense to attach the car seat production to one end of the car assembly line. It may be run by the seat supplier or by the OEM, but anything that can eliminate container trucks with a moving line would result in substantial savings.
In comparison, motorcycle saddles are not complex, even in the most expensive of machines. Even the best motorcycle seats are way low in the pecking order when it comes to BoM cost by component heads in a motorcycle.
Combine everything, and it makes zero sense for a two-wheeler manufacturer to bring seat building in-house. Dedicated seat suppliers have the experience to manufacture seats optimally, and they derive scale efficiency by supplying to multiple OEMs.
Non-Core: What Makes Sense to Integrate?
Apart from the core components – the cells, packs, controllers, and motors- there are also non-core components. These include the frame, suspensions, brakes, seats, wiring harnesses, lamps, plastics, and sheet metal parts. We have discussed seats above, and the same is true for any component class. Specialised suppliers exist in every mature two-wheeler market, and their expertise, efficiency, and cost advantage come from being suppliers to multiple OEMs.
That’s not something that an OEM can replicate.
