BYD and Vertical Integration

They've done well making cars to sell batteries

Intro

I often find insights standing a problem on its head. Most analysis, including my own analysis on SeekingAlpha, views BYD as a car company integrated backwards into battery production. What if instead we think of BYD as a battery company integrated forward into making things that use batteries?

That better explains their range of businesses, which includes assembling cell phones and other portable devices that use batteries, making electrified vehicles, and supplying large-scale energy storage systems.¹

That also changes the strategic analysis: when does it make sense for a manufactuer to integrate forward (downstream)? Standing things on their head:

Does it still make sense for BYD to make cars?

Vertical Integration

Normally I argue that car companies should shun vertical integration into producing parts and components. Of course that begs the question of what is a car company.

A car company undertakes 3 core operations: it is a wholesaler that engineers, assembles, and markets vehicles. For the most part, it does not manufacture the components it assembles. Assembly includes a body shop, a paint shop, and final assembly. Most companies stamp their own body panels adjacent to the body shop, and then paint it before conveying it to final assembly. In addition most companies assemble their own gasoline engines, and many though not all their own transmissions.

There’s a large literature on when to make and when to buy. The trend over time has been to buy, even when a supplier builds a factory dedicated to a single assembly plant of a single car company. Seating is an example.

I won’t get into all the arguments over why firms draw boundaries where; I have a whole book on that.² But basically there are two reasons to do so, a market power (monopoly) story and a transaction cost story. Let me start with the latter. The former is more useful for thinking about the distribution (downstream) end of the industry, the wholesaler piece.

Transaction costs driving (upstream) vertical integration

Henry Ford in the early days couldn’t find machine shops that were willing to invest in specialized tools to make parts for the Model T. After all, he was but one of roughly 300 car companies in the US, and most didn’t last very long. He also wanted a lot of parts, and specialized machines didn’t need skilled machinists (though it then needed machinists to keep the machines running). Most of the machine shops around Detroit were small, and focused on managing skilled workers. So Ford would have had to finance their expansion, and get them to rethink their business model, and do that across many firms. Instead he chose to do it himself.

That logic however was situational. In Japan in the 1950s there were lots of medium sized machine shops, as the pieces of the Japanese war machine tried to redefine their businesses. Ford and GM assembled cars in Japan in the 1930s, plus there were imports, and as the war expanded various firms made trucks for the military. So there were already firms making brakes and radiators and piston rings, to meet demand for replacement parts. Crucially, these small firms all had banking relationships. It was the car companies that had a hard time getting financing. The car companies also faced unions, smaller firms mostly did not. So when for example Toyota came out of its de facto bankruptcy in 1950, it could find suppliers, but it couldn’t find money, and so could barely keep up with building assembly capacity. (Indeed, they turned several failed car companies into contract assemblers to try to stretch their funds further.)³

Europe fell somewhere in between the US and Japan. Again, vertical integration proved contingent, but I won’t try to detail here.

Over time what started at Ford as a solution turned into a problem.

• Vertical integration limited the size of the potential market (TAM). Internal parts operations couldn’t sell to their competitors, limiting economies of scale in production and their ability to share engineering across multiple customers. They only gained knowledge of new approaches at other car companies after they were in the market, so were consistently a couple years behind. Many new innovations (car radios, air conditioning) came from Chrysler or other smaller firms.

• Vertical integration meant there was no incentive to seek markets outside of automotive. Asking for an R&D and marketing budget to explore such markets? Really? And if even if you had items made in your plant immediately salable, asking the 12th Floor for a capex increase to sell components to white goods or the commercial air conditioning wasn’t going to get a good reception.⁴

• Since Ford didn’t shop outside, internal parts operations didn’t have to be competitive in costs, at least as long as they seemed OK when taking GM as a benchmark.

• They were conservative. Over time engineers were shepherded into silos, closures and chassis and so on, and it became challenging to implement ideas that involved other silos. Plants had depreciated equipment, so there was pressure to keep it in use. That applied even more to restructuring – no exec wanted to eliminate their own job.⁵

Now engineers still liked doing new things, and as a smaller firm Ford had more to gain from innovations that did GM. Still, over time Ford’s parts operations became uncompetitive. Then came new demands for safety, for fuel efficiency, and for lower emissions. Worse, the Detroit Three faced new competition in from 1980s by Japanese car companies who squeezed them from below and German car companies who squeezed them from above.

It’s a long saga, but by the 2000s Ford (and GM) undid their vertical integration, most immediately by bundling their parts operations into Visteon and Delphi, and then spinning them off. By that point several hundred European and Japanese suppliers had also set up plants in the US, there was the maquiladora in Mexico, and parts imports from Asia. Ford and GM were thus able to move away from Visteon and Delphi. While some individual plants survive, overall it didn’t end well for workers and their communities, or for the new owners and managers.

So vertical integration is not a no brainer, and the transactions costs that managers might see in one decade can turn into an albatross of technologically stagnant, monopolistic, high-cost internal operations in another. Today a car company in the US may have no choice but to vertically integrate if it wants batteries, but it’s silly to think that such one-off single-plant investments can keep up with rapid innovation.

Running a supply chain is hard, running diverse internal operations is harder.

Market power driving (downstream) vertical integration

I’ll tell the market power story as one of forward integration. If you have market power – in the classroom, “monopolist” because the models are analytically simple – then you of course choose to price your product higher rather than lower. However, if your customer also has market power, then they are doing the same. The result it that the higher price gets amplified, in economics jargon through “double marginalization.” That results in lower sales for both, and lowers profits.⁶

One response is to integrate forward by buying your customer. That, of course, can suffer the problems that I traced for backward integration. In retailing, it leads instead to the use of franchising, where an annual fee makes up for selling to them at a high price. Historically that has been the route used in the auto industry, as independent retailers are more adept at dealing with local markets with their variation in customer needs and tastes, effective advertising channels, pay scales and location.

Ford was the most important innovator in franchising in the 1920s.⁷ By the time of CEO Jacques Nasser the lessons of that era had been forgotten at headquarters. He had Ford buy its Tulsa area dealerships in 1999, paying a large premium as their stores were making good money and they hadn’t been interested in selling. Ford quickly relearned that the efficiencies of corporate standardization didn’t offset the accompanying inflexibilities. Come 2001 and their Tulsa stores were losing money; Ford sold them back to the original owners at a significant discount. In short they were again profitable.⁸

See Glenn Mercer for more on Jac the Knife at Ford, and throughout his substack on the strength of the dealership model and the lack of success of direct sales efforts, such as that of Tesla.

Over time, the forward integration edifice is also prone to fringe competition by players attracted by the high profits. (Clayton Christensen took that well-known story and re branded it, without attribution and with much exaggeration of the empirical case, as “disruption.” I’ve already started drafting that story for a future substack article.)

If you’ve not gotten the message yet, vertical integration is no panacea.

Key though is that…

…market power based (forward) vertical integration only works as long as the downstream customer has market power.

BYD: Batteries and Cars

To date BYD has done quite well integrating forward into making and selling cars. It’s done so, however, with an engineering mindset. In my previous note, and in greater depth at SeekingAlpha, I argue that their days of rapid growth are over. Let me include just two graphs here, of sales of their core Dynasty models (at the bottom) and of the Ocean series. What you see is that, in the hypercompetitive Chinese market, it’s hard to maintain a model’s sales. (What I don’t show here is that this is despite a stream of new variations and model refreshes.) They make great cars, but they’re not great marketers. As one comment at SeekingAlpha claimed, the founder, a metallurgical engineer, has all the charisma of the periodic table…they sell on price, not brand strength.

In the early days, there weren’t many EV makers, and in 2022 BYD held over 44% of the engineering-intensive PHEV subset.⁹ That is, they had market power. Others, however, entered the market. In 2025Q3 they only had a 28% share, and in October 2025 it was down to 26%.

From another direction, in 2022, Chinese car companies launched 232 new models (including full refreshes), of which only 31 were PHEVs. So far in 2025 they’ve come out with over 80 (while pure ICE launches fell from 131 in 2022 to 27 so far this year).¹⁰

BYD now has full coverage of the market, from subcompacts to full-sized SUVs, so the rapid growth that came from expanding their lineup is at end. With many models now 3 years old, their investment now has to be devoted to renewing existing product. Arguably, as per the PHEV data, they no longer have market power and the profits that came with it (echoed in their quarterly financials). Ahead is the long slog of improving their branding, and pushing exports. Both will take resources, and hold down profits for at least the next couple years.

Implications

All this suggests that BYD’s forward integration into cars no longer makes sense. It impedes their battery sales to other car companies. It doesn’t boost combined profits. It won’t produce growth, indeed if they are compelled by tariffs to make batteries in Europe and elsewhere, it will result in having to spend money on low-volume plants that will not achieve economies of scale and will face higher input costs because they their battery supply chain remains in China.

That has one clear implication. Forward vertical integration no longer makes sense, so

Cars used to be a plus, but now are a drag on the firm as a whole. BYD should therefore spin off its battery business from its car business.

Now cars are a consumer product, but batteries are an intermediate product where branding is irrelevant – I have many battery-operated products, but neither know nor care who makes those batteries. So the car business should keep the BYD name, and the battery side should take a new name, with its own public listing.

My hunch is that both the “handset” business and the energy storage business have market power, but the same logic applies: at some point they, too, should be made independent.

Strategic and management and technology concerns reinforce that. Within the current BYD structure, the R&D groups probably have little overlap.

• The chemical and process engineering for batteries is independent from developing new cars, and new cars from PHEV drivetrains. Keeping process yields high and staying abreast with rapidly evolving chemistries is the name of the game. Sales, well, most users would like to have more than one source, so becoming adept at meeting different packaging needs, and providing sales engineering for your packing, is central. Then there is a very different supply chain of specialty chemical firms and metal refiners, and access to raw minerals. That’s not something in the automotive industry’s comfort zone.

• Handsets are surely labor-intensive, require a lot of attention to automating quality control, and face the challenge of working with a lower-wage laborforce with high turnover. Those operations contract with cell phone companies, and so have a very short list of customers; maintaining those partnerships is the other strategic challenge.

• In contrast, energy storage is marketed to electric utilities, and “turnkey” projects have a different timeline and require a local sales support structure in multiple countries. That doesn’t overlap with what’s needed to sell the latest battery cell (or nowadays, “blade” that replaces multiple cells).

We can take it one step further, noting that diesel engine manufacturers are generally independent of end users. In fact, Geely has spun off its PHEV/hybrid drivetrain operation as Aurobay, and now has sales (and investments) from other carmarkers. To my knowledge, BYD doesn’t market their PHEV drivetrains – they are no longer capacity constrained, as BYD had 2024Q4 sales of 1.25 million vehicles, but only 0.93 million in 2025Q3.

So … disintegrate!

Arrows below indicate when BYD refreshed models. It’s not a complete list, but BYD has been around long enough that they are having to devote resources to redesigning existing models. I don’t add such detail for the older Dynasty series, there would be too many arrows.

Addendum

FinDreams is BYD’s wholly-owned battery subsidiary, and I presume the seller to outside automotive customers. If I were a car company, buying from a 100% subsidiary would not instill confidence that BYD would never skew terms to my disadvantage.

Note that while FinDreams was set up in 2019, the timeline on their website starts with initial research into LFP cells in 1996, with cell phone customers Motorola from 2000 and Nokia from 2002. At that point they claimed a 30% global market share for mobile phone batteries. They entered the laptop battery market in 2013, and started production at their first plant outside China (in Thailand) in 2024.¹¹

Notes

1. I initially wrote small but growting but the asked asked Chrome “where had byd had energy storage projects.” Thta generated an AI search at that over the past 17 years they had installed 75GWh across 350 projects. I then asked about Tesla, which answered 31.4 GWh in 2024 and 23.5 GWh 2025Q1-Q3/ So I then put in a date, and got BYD installations at 40GWh in 2024. I then asked about CATL, and was told 110GWh in 2024. So it looks like CATL holds a solid lead, with BYD in second and Tesla third.

2. Michael Smitka, Competitive Ties: Subcontracting in the Japanese Auto Industry (Columbia University Press, 1991).

3. My Zotero bibliographic database is corrupt. Until I rebuild it I can’t readily provide references to the business history literature that I last looked at 15 years ago, there are a couple articles specific to Toyota’s contract assembly. Now I do have a really good Japanese-language business history that covers the 1945-1965 era, but couldn’t find it. Likely it’s in my unheated garage, on a wall behind ladders and tools. Yeh, crocodile tears. Email me if you really want to know.

4. Where most of the executives were – Ford is buididng a new HQ. They’d gather together every Thursday in the Thunderbird Room for the senior management meeting, where (through alumni connections) one or another senior person met with my students. The most memorable was almost 2 hours with CFO Lewis Booth, who walked them through the internal discussions that let Ford avoid Chapter 11 during the Great Recession. Booth had just finished briefing Mr. Ford ahead of their annual meeting in Delaware the next day, and with all the preparations finished, was unwinding with us. No one asked for a copy of his slide deck, my students sensed how special and honest that briefing was.

5. My knowledge of engineering silos is specific to GM; but Ford was smaller so people may have moved around more and had personal experience and strong ties to other groups. See James Womack et al., The Machine That Changed the World (1990) and Kim Clark and Takahiro Fujimoto, Product Development Performance: Strategy, Organization, and Management in the World Auto Industry, Product Development Performance: Strategy, Organization, and Management in the World Auto Industry (1991).

Closing a plant faced pushback, not just at the local level but also with upper management, as salaries and status were tied to headcount. With fewer subordinates, a middle manager might no longer qualify for a bonus and leased car.

6. If you want the graphs, google “double marginalization” and you’ll find plenty of explications. The article on Wikipedia is fine, but uses equations, not graphs.

7. See Thomas Dicke’s Franchising in America: The Development of a Business Method, 1840-1980 (University of North Carolina Press, 1992), which has a long chapter on Ford as the core innovator. On substack, see Glenn Mercer’s CarCharts, and

8. See David Ruggles’s blogs at AutosandEconomics. David is a life-long dealership guy, and was originally from Oklahoma. By the time of the Lee Iacocca and the 1979 Chrysler bailout, he was running a Chrysler dealership in Chicago. It was near to a TV station that made his store the go-to location when they needed a segment, and so periodically his face appeared in the national news. Later he ran luxury stores and then turned to consulting, teaching credit unions how to do auto financing. David is now fully retired.

On substack and still active, see Glenn Mercer’s CarCharts, Glenn’s career has been in consulting, with 2 decades at McKinsey and the last 2 decades independent. Back in the days he was involved in the MIT International Motor Vehicle Program, sat on corporate boards, the big stuff, not like yours truly who’s based in rural Virginia and taught all sorts of undergraduate economics classes.

9. In my database I lump together sales data of models that come in both EV and PHEV versions as NEVs, so I can’t get a “clean” number – I can either use PHEVs plus EVs, which gives me an over count of PHEVs, or I can use just PHEVs, which gives me an undercount. Because of BYD’s naming conventions – DM-i models are PHEVs – I can mostly separate out EV sales from PHEV sales for them, but they still have NEV models that include both.

10. When I track new models, I count one that comes as both an EV and a PHEV as 0.5 of each. That way my total number of new models matches my ICE + PHEV + EV total, but I failed to add a column to my spreadsheet to count the number with both. I ought to go back and correct that, but it’s a lot of weekly reports to rescan…

11. The website called it an “overseas” plant, but while they don’t share a border, you can go overland to reach Thailand through either Myanmar or Laos, or via Vietnam and then via Cambodia or Laos. Google maps prefers the latter, giving a road trip of 2,650 km from BYD’s headquarters in Shenzhen to Bangkok of 35-42 hours, depending on time of day as the route passes across Guangzhou, Nanning and Hanoi.