Seres Group: The Huawei Handshake and the Twenty-Billion-Dollar Auto Turnaround

I. Introduction & Episode Roadmap

On the morning of July 13, 2026, the order book for 赛力斯集团 Seres Group Co., Ltd. opened on the 上海证券交易所 Shanghai Stock Exchange and simply stopped. Limit down. No trades. The company that had, eighteen months earlier, been the most celebrated turnaround story in Chinese industry — a microvan assembler that had become the best-selling luxury car brand in the world's largest auto market — had just warned that it would lose between RMB 1.5 billion and RMB 1.8 billion in the first half of 2026, against a RMB 2.94 billion profit in the same period a year before.³³ From a peak of RMB 173.55 per share in September 2025, the stock had fallen close to seventy percent in under ten months, erasing more than RMB 200 billion of market value.³³

Here is what makes that collapse analytically interesting rather than merely dramatic: it happened while sales were rising. Deliveries of the 问界 AITO brand grew 10.2% year-on-year in the first half of 2026.³⁰ More cars, less money. That single fact is the doorway into the whole Seres question.

Because the story most people tell about Seres is a fairy tale, and fairy tales make bad investment theses. The tale goes like this: a provincial Chongqing metal-basher that made springs for washing machines somehow shook hands with 华为 Huawei, and the handshake turned it into a luxury automaker with RMB 165.05 billion of 2025 revenue and RMB 5.96 billion of net profit — the fastest revenue expansion in modern automotive history.¹ The numbers are real. The 2024 revenue growth rate of 305% is real.¹⁸ What the fairy tale leaves out is the price of the handshake, who sets that price, and what happens to the arithmetic when the volume leader shifts from a RMB 500,000 SUV to a RMB 300,000 one.

So the twenty-billion-dollar question — roughly the scale of Seres' annual revenue in US dollar terms — is this. Is Seres a genuine automaker that has bought itself durable access to the best consumer-technology stack in mobility? Or is it a contract manufacturer with a stock ticker: a body shop that assembles vehicles designed, branded, marketed, and sold by someone else, keeps a thin residual margin, and carries the full weight of the capital, the inventory, the tooling, and the warranty?

Founder 张兴海 Zhang Xinghai would say the first. The bears would say the second. The honest answer, as we will see, is that the evidence supports something in between — and that the direction the answer moves in over the next two years is the entire investment case.

Here is the road we will travel. We begin in the industrial hill country outside Chongqing in 1986, with three brothers, RMB 8,000, and a factory that made springs for washing machines. We follow that business through motorcycle shock absorbers into a state-private joint venture that made some of the cheapest four-wheeled transport on earth. Then comes the detour that nearly killed the company: a Silicon Valley electric-vehicle adventure that burned years and billions and shipped almost nothing. Out of that wreckage comes the decision that defines the company — saying yes to Huawei on terms that China's largest automaker publicly called a surrender of the corporate soul.

From there the story turns commercial. We will trace the AITO brand from its embarrassing first attempt through the redesigned M7 that Huawei's own executive described as a resurrection, into the M9 flagship that outsold the German luxury establishment. We will audit the capital allocation: RMB 2.5 billion for a trademark portfolio, RMB 11.5 billion for a tenth of Huawei's automotive-technology arm, HK$14.3 billion raised in Hong Kong. We will size the segments, stress-test management's credibility against their own record, run the business through Hamilton Helmer's 7 Powers and Porter's Five Forces, and finish with the three metrics that will actually settle the argument.

One framing device is worth establishing at the outset, because it recurs at every stage. In the century-old automobile industry, value accrued to whoever controlled the hardest thing to replicate — first the engine, then the platform, then the brand, then the dealer network. In the software-defined vehicle, the hardest thing to replicate has migrated again, to the operating system, the driver-assistance stack, and the consumer relationship that sits on top of both. Seres' entire corporate strategy is a wager that when the scarce layer moves and you do not own it, the rational response is to rent it rather than to build a worse version of it or pretend it does not matter.

That wager is not obviously wrong. Most of the world's most profitable manufacturers rent something — Foxconn rents Apple's demand, Taiwanese fabless designers rent TSMC's process. But renting the customer-facing layer is a different order of dependence from renting a factory, because the landlord can invite your competitors into the same building. Which, as it happens, is precisely what Huawei did.

The through-line is a question that applies far beyond one Chinese carmaker: what is a manufacturer worth when someone else owns the customer?

↑ Back to Top

II. The Chongqing Roots: Springs, Shock Absorbers & Microvans

Fenghuang Town sits in the hills of what was then Ba County, on the outskirts of Chongqing — a city of staircases and river fog that Mao's planners had stuffed with defense factories and machine shops during the Third Front industrialization. In September 1986, three brothers named Zhang — Xinghai, Xingming, and Xingli — pooled their entire savings to raise RMB 8,000 and opened the Ba County Fenghuang Electrical Spring Factory.³ Zhang Xinghai was twenty-three. RMB 8,000 in 1986 was several years of a factory worker's wages, and it bought them exactly one thing: the right to make a coiled piece of wire slightly better and slightly cheaper than anyone else.

They made square-wire garter springs for washing machines. It is difficult to imagine a less glamorous product, and that turns out to be the point. Within two to three years, the brothers held roughly ninety percent of the Chinese market for that specific component.³ Not because of technology — because of relentless cost discipline in a category too small and too dull for anyone with better options to bother contesting. This is the founding personality trait of the business, and it recurs at every stage of the story: Seres has consistently won by being willing to take the unglamorous, low-status position in a value chain that other people find beneath them.

The spring factory became a shock absorber factory, and by 1995 the operation had been restructured into Chongqing Yu'an Group, aimed squarely at the motorcycle and automotive suspension market.³ Chongqing in the 1990s was to motorcycles what Shenzhen was later to phones — a dense cluster of assemblers and component makers producing tens of millions of cheap two-wheelers for domestic buyers and export markets across Southeast Asia and Africa. Yu'an supplied that cluster. It learned to run high-volume, low-margin, high-reliability manufacturing lines where a one-yuan cost saving mattered and a quality escape could kill a customer relationship. Again: unglamorous, and again, decisive later.

The pivot from components to whole vehicles came in 2003, and it came through a door that only a well-connected private Chinese industrialist could open. Chongqing Sokon entered a fifty-fifty joint venture with 东风汽车 Dongfeng Motor Corporation, one of the great state-owned automotive groups, creating 东风小康 Dongfeng Sokon — DFSK.⁴ The structure was elegant for both sides. Dongfeng brought the manufacturing licence, the state imprimatur, and a national distribution reach that a private Chongqing family firm could not have assembled. The Zhang family brought cost engineering and the willingness to operate at margins a state enterprise would find humiliating.

What DFSK built were microvans and light trucks: boxy, upright, four-to-seven-seat vehicles priced in the four-to-seven-thousand-dollar band, sold into tier-three and tier-four cities and the rural logistics economy. These were not cars in the Western consumer sense. They were capital equipment for small businesses — the vehicle a rural entrepreneur bought to haul vegetables to market, carry construction materials, or run a village delivery route. The segment was enormous, brutally price-competitive, and dominated by 五菱 Wuling. DFSK fought its way into the top ranks of it.

It is worth pausing on what that business model demanded, because the demands are the opposite of what a premium manufacturer faces. At a four-thousand-dollar selling price, there is no room for brand premium, no room for dealer margin theatrics, and no room for engineering elegance. The entire profit sits in the gap between a fiercely policed bill of materials and a price the market sets for you. Gross margins in the mid-single-digits to low-teens were normal. Volume was the only lever that mattered, and volume required flawless supplier coordination across hundreds of low-cost Chinese vendors, plants that could run at high utilization on thin shift premiums, and a tolerance for competing on nothing but cost against a rival — Wuling, backed by both SAIC and General Motors — that had every structural advantage.

There is a straight line from those microvans to the AITO M9. Building a vehicle that must sell profitably at RMB 40,000 forces an organization to internalize stamping, welding, tooling, supplier negotiation, and line-balancing at a level of granularity that a premium manufacturer, cushioned by price, can afford to outsource.

Seres did not learn to make luxury cars by studying Mercedes. It learned to make cars by making the cheapest ones in the country and surviving.

The capital markets chapter opened on June 15, 2016, when Chongqing Sokon Industry Group listed on the Shanghai Stock Exchange under the ticker 601127. The IPO was modest to the point of being unremarkable: 142.5 million shares at RMB 5.81, raising RMB 828 million gross and roughly RMB 738 million net, earmarked for automotive components and engine-parts production lines.⁵ This is a useful anchor. The company that would later raise HK$14.3 billion in Hong Kong entered public markets with less than a fifth of a billion US dollars, a commercial-van business generating single-digit net margins, and no meaningful exposure to electric passenger vehicles.

For investors, the takeaway from this era is not nostalgia. It is that Seres' genuine, self-owned capability — the thing it possesses independent of any partner — is high-volume, low-cost vehicle manufacturing and supply-chain execution. Everything premium about the company today was acquired, licensed, or borrowed. Which raises the obvious question: what happened when this cost-obsessed component company first tried to buy its way upmarket on its own?

The answer is the most expensive mistake in its history.

↑ Back to Top

III. The Silicon Valley Detour & The $500M US EV Gamble

Picture Santa Clara, California, in 2017. A Chinese microvan manufacturer with barely a year of listed history has opened an office in the heart of Silicon Valley, hired away one of the most recognizable names in electric vehicles, and started telling American reporters it will build luxury electric SUVs in the American Midwest. If you had been handed that description without a name attached, you would have assumed it was a fraud or a fantasy. It was neither. It was Sokon, and it was entirely sincere.

The strategic logic was sound even if the execution was not. Chinese microvan demand was structurally rolling over as rural incomes shifted buyers toward passenger cars, and Beijing was pouring subsidies and mandates into New Energy Vehicles. A company whose entire franchise sat in cheap internal-combustion commercial vehicles was staring at a slow death. Zhang Xinghai's read — get into electric passenger vehicles, get in early, and get in at the premium end where the margins are — was the correct read. What he got wrong was where to do it.

In October 2017, SF Motors, the Sokon-backed US entity, announced the acquisition of InEVit, an electric-vehicle battery modularization startup founded by Martin Eberhard.⁶ Eberhard was not a marginal figure. He co-founded Tesla and served as its first chief executive, and he brought patented battery module design and chassis architecture work into the deal, joining SF Motors as chief science officer. For a Chongqing spring-maker, this was an extraordinary talent coup, and it is worth understanding what Sokon believed it was buying: not just batteries, but legitimacy. The message to Chinese regulators, investors, and consumers was that this was a serious technology company, validated by the man who started Tesla.

Three weeks later, on November 2, 2017, SF Motors closed the purchase of AM General's commercial assembly plant in Mishawaka, Indiana — the factory that had built the Hummer H2 — preserving roughly 430 American jobs.⁷ By June 2018 the company had committed to invest around $160 million in the facility to build electric SUVs.⁸ Prototypes were unveiled: the SF5 and SF7, positioned against the Tesla Model X and Model Y.

Read the press releases from that period and the ambition is dizzying. A Chinese company that had never sold a passenger car outside China was promising American-built premium electric SUVs, protected-battery architecture, and driver-assistance features, with production timed for the end of the decade. There was a plant, a chief science officer with an unimpeachable pedigree, a Californian address, and state economic-development officials publicly celebrating the return of manufacturing jobs to northern Indiana. Everything was in place except the two things that actually sell premium cars: a software organization capable of building an experience customers would prefer, and a reason for an American buyer to choose an unknown badge over a Tesla.

And then, essentially, nothing.

In July 2019, SF Motors' new chief executive James Taylor read a letter to employees announcing layoffs. Roughly thirty of the eighty-one workers at Mishawaka lost their jobs, and US development of the SF5 was put on hold, with the company citing a weakening Chinese auto market and the escalating US-China trade conflict.⁹ The Indiana plant was subsequently sold to Electric Last Mile, a Michigan electric van startup.¹⁰ Not a single mass-production retail vehicle was delivered to an American customer.

The scale of the burn deserves to be stated plainly, because it is the single most important piece of disconfirming evidence against any claim that this management team is a naturally gifted capital allocator. Sokon spent years and billions of renminbi on US research and development, Silicon Valley overhead, a factory, tooling, and a marquee executive hire, and converted it into an asset sale and a write-down. Look at what happened to the parent's financials as the bill came due. In 2018 the group earned RMB 95 million of net profit on RMB 20.2 billion of revenue; in 2019, RMB 67 million on RMB 18.1 billion.³⁷ Those are rounding-error profits on a substantial revenue base — an operating business already being consumed by the EV programme.

Then the losses began in earnest: RMB 1.73 billion in 2020, RMB 1.82 billion in 2021, RMB 3.83 billion in 2022, and RMB 2.45 billion in 2023 — four consecutive loss years totalling RMB 9.835 billion.³⁷¹⁷

The benchmarking is unkind. 蔚来 NIO and 小鹏 XPeng were burning comparable or larger sums in the same window, but they were building native software organizations, proprietary user interfaces, direct-to-consumer retail networks, and brands that Chinese premium buyers recognized. Sokon was buying hardware assets and a famous résumé. It had no software culture, no brand equity in North America, no dealer or service network there, and no plausible path to distribution. The InEVit acquisition brought battery packaging know-how into a company whose deficit was never battery packaging — it was everything the customer actually touches.

By 2019, US manufacturing plans were shelved and the vehicle assets were repatriated to China under the 赛力斯 Seres name. Management would later frame this as decisive loss-cutting, and there is something to that: they did stop, rather than doubling down into insolvency. But the honest characterization is narrower. Seres did not identify the error early; it identified the error after the money was gone and the trade war made continuation impossible. Investors evaluating today's capital allocation — RMB 11.5 billion into a single minority stake, for instance — should weigh this record. The company's demonstrated skill is recognizing when a bet has already failed, not avoiding the bet.

What Seres emerged with was an electric powertrain and vehicle-platform capability, a heavily impaired balance sheet, a premium-vehicle plant in Chongqing with nothing profitable to build in it, and a founder who had just learned, expensively, that he could not buy his way into the consumer-facing layer of the automobile.

Six hundred miles east, a telecommunications company was learning the opposite lesson.

↑ Back to Top

IV. The Faustian Bargain: Handing the Keys to Huawei

In May 2019, the US Commerce Department added Huawei to the Entity List. What followed was the fastest destruction of a consumer-electronics franchise in modern history: cut off from advanced chips and from Google's Android services, Huawei's smartphone business — the second-largest in the world — collapsed. The company had one of the deepest hardware and software engineering benches on the planet and, suddenly, a shrinking product to point it at.

余承东 Richard Yu (Yu Chengdong), the combative and famously quotable head of Huawei's Consumer Business Group, needed a new arena that used the same muscles: chips, operating systems, sensors, industrial design, retail, and marketing. He found it in the car. Not in building cars — Huawei has insisted, repeatedly and to the point of exhaustion, that it does not build cars — but in supplying and selling everything about a car that is not the metal.

Yu is a specific kind of executive and the specificity matters to this story. He built Huawei's phone business from an obscure white-label operation into a global number two by making enormous public claims, missing some of them, and hitting enough of the rest that the claims became self-fulfilling. He is the executive who stands on stage and tells you the product is the best in the world before it has shipped. In a mature industry that treats forecasts as contractual, this style is a liability. In Chinese consumer electronics, where attention is the scarcest input, it was a weapon — and when he turned it on automobiles, an industry whose product launches had for a century been decorous affairs run by marketing departments, the effect on order books was immediate and measurable.

Huawei structured its automotive ambitions into tiers. At the bottom was ordinary component supply. In the middle sat the Huawei Inside model, where the carmaker integrates Huawei's full-stack driving and cockpit systems but keeps the brand and the showroom. At the top sat the model that would define Seres: 智选车 Smart Selection, later formalized as the Harmony Intelligent Mobility Alliance, or HIMA. Under Smart Selection, Huawei participates in defining the vehicle itself — how it looks, what it does, what the software feels like. It supplies the HarmonyOS cockpit, the ADS driver-assistance stack, and DriveONE electric powertrain components. And critically, it sells the car through its own nationwide retail network: the same gleaming stores where hundreds of millions of Chinese consumers buy phones and tablets.

For a traditional automaker, this proposition is close to unthinkable. Strip out product definition, user experience, software, marketing, and retail, and what remains for the manufacturer is stamping, welding, painting, assembly, quality control, and warranty liability. You keep the capital intensity and the recall risk; you surrender the customer relationship and the brand.

上海汽车集团 SAIC Motor chairman Chen Hong said the quiet part out loud at SAIC's shareholder meeting in mid-2021. Asked about working with Huawei on autonomous driving, he described the arrangement as one in which the technology partner "becomes the soul and SAIC becomes the body," and concluded that "we want to take the soul into our own hands."¹¹ The 灵魂论 "soul theory" instantly became the organizing metaphor for the entire Chinese auto industry's anxiety about Huawei. It also, in retrospect, became a monument to how expensive corporate pride can be — SAIC would eventually join HIMA itself with the Shangjie brand, four years later and from a position of far greater weakness.⁴¹

There is a second, less discussed asymmetry embedded in the Smart Selection structure, and it explains why Huawei wanted it. Huawei's constraint after the sanctions was not talent or technology; it was capital efficiency and political exposure. Owning car plants would have made it an automaker in fact, invited fresh scrutiny, tied up tens of billions of renminbi in fixed assets, and — most damagingly — turned every existing or prospective Chinese automotive customer into a direct competitor. By defining, branding, and selling cars that someone else builds, Huawei captures the value of the consumer relationship while keeping its balance sheet clean and its supplier pitch to the rest of the industry technically honest. It is a structure designed to maximize Huawei's return per unit of Huawei's capital. Seres supplies the capital.

Zhang Xinghai's calculation was different, and it was different because his position was different. SAIC had brands, dealers, and a century-equivalent of accumulated consumer trust to protect. Seres had a Silicon Valley crater, four years of losses ahead of it, an underutilized premium plant, and no consumer brand worth defending.

The soul question is only agonizing if you have a soul the market values. Seres did not. What it had was capacity, cost discipline, and desperation — which, from Huawei's perspective, made it the ideal first partner: capable enough to build, weak enough to accept the terms.

The first attempt was not a success, and it is important to say so, because the AITO story is usually told as though it began in triumph. On April 21, 2021, the Seres SF5 Huawei Smart Selection went on sale at RMB 216,800 to RMB 246,800, becoming the first car ever sold inside Huawei's flagship retail stores.¹² Orders exceeded 3,000 within two days, and Richard Yu talked publicly about vast volume ambitions.¹² Reality intervened. By the end of November 2021, cumulative SF5 sales had reached fewer than 7,000 units, and owners were publicly complaining about a list of quality and specification problems.¹³

That failure is analytically valuable in a way the later successes are not. It demonstrates that Huawei's retail footprint and software, applied to a vehicle that was fundamentally a warmed-over Seres product, did not sell. The traffic in the stores was necessary but not sufficient. What was missing was a vehicle designed from the ground up around the Huawei experience — which is precisely what the two companies went away and built.

The division of labour that emerged was explicit and remains the load-bearing structure of the whole enterprise. Huawei defines the vehicle, owns the interior and interface design language, writes the software, supplies the intelligent hardware, runs the marketing, and sells the car. Seres engineers the platform, invests in the tooling and the plants, procures the mechanical supply chain, builds the vehicle, and stands behind it. Huawei carries almost no fixed asset risk. Seres carries almost all of it.

Whether that is a bargain or a Faustian bargain depends entirely on the split of the economics — a number that would remain opaque to public investors for another four years, until a Hong Kong listing prospectus forced it into the daylight.

↑ Back to Top

V. The `问界 AITO` Inflection & Financial Turnaround

The AITO brand launched in December 2021 — the Chinese name 问界 meaning, roughly, "to inquire into the boundary," the English an acronym for Adding Intelligence to Auto. The first product, the M5, was an extended-range electric SUV: a battery-electric drivetrain with a small petrol engine on board acting purely as a generator, never driving the wheels. For readers unfamiliar with the architecture, think of it as an electric car that carries its own portable charger. It delivers the smooth, quiet, instant-torque driving experience of an EV while eliminating range anxiety entirely, at the cost of hauling around an engine most of the time. In a country where charging infrastructure was still uneven outside major cities, this was a shrewd product choice, and Seres would go on to hold the largest share of China's range-extended segment.¹

The M5 earned genuine praise for its HarmonyOS cockpit — the seamlessness of walking from a Huawei phone into a Huawei-powered car was a real and novel consumer experience — but it landed into a fight it could not win. 理想汽车 Li Auto had already built a commanding position in exactly this segment with a family-oriented product line and a devoted following, and it had spent years learning what Chinese families with two children and a set of grandparents actually want from a vehicle. AITO arrived with better software and a weaker product proposition.

Through 2022 and into 2023, Seres kept losing money at scale, and by early 2024 the company was guiding to a fourth consecutive annual loss even as it told investors that the M7 and the forthcoming M9 were paving the road back to profitability.¹⁶ It is worth registering how thin the ice was at that moment. A company that has lost money for four straight years, whose flagship partnership has produced one commercial failure and one underwhelming launch, does not get many more attempts. The alternative history in which Huawei quietly reallocates the Smart Selection franchise to a healthier partner in 2023 is entirely plausible.

The turn came on September 12, 2023, and it came from a decision that looks obvious in hindsight and was not obvious at all at the time: rather than launching a new nameplate, Seres and Huawei took the M7 — a model that had underperformed — and comprehensively re-engineered it. Structural safety was upgraded, the interior was redone, Huawei's newer ADS driver-assistance stack was installed, and Richard Yu cut the entry price by RMB 40,000, bringing a five-seat variant in at RMB 249,800.¹⁴

What happened next was the most important commercial event in the company's history. Firm orders — deposits, not expressions of interest — passed 100,000 in under three months and reached roughly 130,000 within four months of launch.¹⁴¹⁵ Seres pushed M7 production toward 700 units a day and lifted annual capacity toward 200,000 units, with the constraint on sales becoming manufacturing throughput rather than demand.¹⁴ Yu, never a man to undersell, called it "a comeback from the dead."¹⁴

Strip away the theatre and consider what the M7 relaunch actually proved. It proved that the combination — a competently engineered vehicle, Huawei's software, Huawei's stores, and an aggressive price — could generate demand that Seres alone had never come close to producing. It also proved something less comfortable: that the demand was highly sensitive to price and to product freshness. A RMB 40,000 price cut was part of the trigger. Investors should hold that thought.

Then came the flagship. The M9, a full-size luxury SUV, moved Seres into a price band no Chinese brand had credibly occupied. It surpassed 100,000 firm orders within six months of going on sale.¹⁹ By 2025 it was delivering more than 110,000 units a year and had led its segment for two consecutive years, and by 2026 Huawei was publicly claiming the M9 dominated Chinese luxury sales above RMB 500,000 — territory that had belonged, unchallenged for two decades, to the BMW X5, the Mercedes-Benz GLE, and the Audi Q7.¹⁴⁶ A mid-size sibling, the M8, launched on April 16, 2025 from RMB 359,800 and gathered more than 80,000 orders within a month, going on to become the group's volume leader at over 150,000 deliveries in 2025.²⁰²¹¹

The financial transformation that followed is genuinely without precedent at this scale. Revenue went from RMB 35.84 billion in 2023 to RMB 145.18 billion in 2024 — growth of 305% — and the group swung from a RMB 2.45 billion net loss to RMB 5.95 billion of net profit, its first annual profit since 2022 and the end of that four-year losing streak.¹⁷¹⁸³⁷ NEV sales rose 182.84% to 426,900 units.¹⁷ In 2025, revenue reached RMB 165.05 billion, up 13.7%, with net profit of RMB 5.96 billion, an NEV gross margin of 28.8%, and operating cash flow of RMB 28.91 billion.¹²

Now the analysis, because those numbers require unpacking rather than applause.

First, the 28.8% gross margin is real but flattering, because of where the Huawei costs sit. In 2025 the group booked RMB 23.43 billion of selling expenses — 14.2% of revenue — against RMB 48.1 billion of gross profit.³⁷ A very large share of that selling expense is the channel and marketing fee paid to Huawei for access to its stores and brand. In an ordinary automaker, distribution economics show up partly in the wholesale-to-retail spread; here they sit below the gross margin line.

The consequence is that a headline gross margin that looks like Porsche translates into a 2025 operating margin of about 4.6% and a net margin around 3.6%.³⁷³³ Seres is not a high-margin luxury business. It is a high-gross-margin, high-distribution-cost business whose bottom line resembles a competent mass-market manufacturer.

Second, the growth deceleration is stark and was visible before the 2026 stumble. After 305% in 2024, revenue grew 13.7% in 2025, and unit NEV sales grew 10.63%.² By the third quarter of 2025, revenue was up 15.75% but quarterly net profit was already down 1.74% year-on-year.²⁹ The turnaround had substantially completed itself by mid-2025; what came after was an ordinary automaker with an extraordinary partner.

Third — and this is the mechanism that broke in 2026 — the profit engine was mix, not volume. In the first half of 2025, revenue actually fell 4.06% year-on-year to RMB 62.4 billion on total vehicle sales of 198,600 units, while net profit rose 81% to RMB 2.94 billion, because the average transaction price per vehicle exceeded RMB 400,000.³⁸⁴⁵ Fewer, richer cars. That lever is powerful and it is reversible, which is exactly what happened.

The product cadence through this period also tells you something about where the demand-generation power actually sits. When Seres and Huawei began pre-sales of an updated M7 in September 2025, the model gathered over 100,000 pre-orders within a single hour.⁴³ No traditional automaker in China generates that kind of instantaneous response, and no Seres product did before Huawei. The counterpoint is that a pre-order collected in an hour is a low-commitment signal, and the company's revenue trajectory over the following twelve months shows that converting that attention into premium-priced deliveries is a different problem from generating it.

The corporate identity caught up with the transformation in July 2022, when Chongqing Sokon formally renamed itself Seres Group, retiring the name it had carried through microvans and the American misadventure.³ With the brand ascendant and the balance sheet repaired, management turned to a different problem — one that had been sitting uncomfortably in the background since the first day of the partnership. Seres had built a hugely valuable consumer brand. It did not own it.

↑ Back to Top

VI. Capital Allocation & Ecosystem Locking: AITO Trademarks & Shenzhen Yinwang

Every analyst covering Seres in 2023 asked some version of the same question: what stops Huawei from taking the AITO name to a different manufacturer? It was not paranoia. Huawei owned the 问界 trademarks. Seres owned the factories. In a dispute, one of those assets is portable and the other is not.

The answer arrived on July 3, 2024, and it was not the answer anyone expected. A Seres subsidiary agreed to acquire from Huawei 919 registered or pending AITO trademarks and 44 design patents for RMB 2.5 billion.²² The striking detail was the valuation: an appraisal put the market value of the portfolio at RMB 10.23 billion as of May 31, 2024.²² Seres paid roughly a quarter of appraised value.

Why would Huawei sell a ten-billion-yuan asset for two and a half billion? Richard Yu's explanation was regulatory: Chinese rules require that the entity owning a vehicle brand and the entity manufacturing the vehicle be the same, and Huawei's absolute public commitment to not building cars made continued brand ownership untenable.²³ Huawei stated it would continue supporting Seres in selling AITO vehicles.²²

Investors should read this transaction with some care rather than treating the discount as a windfall. What Seres bought was legal title to a name, and legal title removes a specific, acute risk — waking up to find AITO on someone else's assembly line. What it did not buy is the thing that makes the name valuable. If Huawei withdrew its software, its retail channel, and Richard Yu's promotional machinery, Seres would own 919 trademarks attached to a brand consumers valued precisely because of what had been withdrawn. The purchase converted an existential risk into a manageable one. It did not create independence. The discount to appraised value is best understood not as Huawei's generosity but as an accurate reflection of what the trademarks are worth separated from Huawei — which is far less than RMB 10.23 billion.

The second transaction was larger and more consequential. In January 2024 Huawei spun its Intelligent Automotive Solution business into a standalone entity, 深圳引望智能技术 Shenzhen Yinwang Intelligent Technology Co., Ltd.²⁵ In August 2024, Seres announced it would pay RMB 11.5 billion in cash for a 10% stake, valuing Yinwang at RMB 115 billion, or roughly $16 billion.²⁴[^27] 长安汽车 Changan Automobile's Avatr unit took an identical 10% on identical terms, with both deals closing on February 28, 2025 and both companies' chief executives joining the Yinwang board, chaired by Huawei's 徐直军 Xu Zhijun.²⁵ Seres completed payment of the final RMB 3.45 billion tranche in the autumn of 2025.²⁶

It is worth explaining plainly what Yinwang actually sells, because the technology is often described in terms that obscure the business. Three product families matter. The first is the intelligent cockpit: the operating system and screens that run everything a driver touches, which in Huawei's case is HarmonyOS — the same software family that runs its phones, so that calls, navigation, music, and calendars move between pocket and dashboard without configuration. The second is the driver-assistance stack, ADS, which combines cameras, radar, and lidar with the processing hardware and software to handle highway and increasingly urban driving under supervision. Lidar, for readers unfamiliar with it, is a laser-based sensor that builds a precise three-dimensional map of the surroundings and works in conditions that defeat cameras; the current M9 generation carries a six-lidar array.⁴⁷

The third is the electric drive system, DriveONE, which packages motor, inverter, and control electronics into a single unit. Together these represent most of what a modern buyer experiences as the "smartness" of a car — and none of it is Seres' intellectual property.

The strategic rationale is genuinely strong on paper. Yinwang sells intelligent driving and cockpit systems to the entire Chinese industry, not just to HIMA partners. Huawei's automotive unit grew revenue 474% to RMB 26.35 billion in 2024 and turned its first profit.²⁵ A 10% holder participates in the economics of Huawei's technology being adopted by competitors, gains board-level visibility into the roadmap, and — the part management emphasizes most — secures priority access to the newest driver-assistance silicon and software.

Now apply the historical falsification test, because this is the largest single capital deployment in the company's history and it deserves the same scrutiny the American adventure earned. The relevant precedent is the InEVit purchase: Seres has previously paid for technology access and famous names and received little of commercial value. The differences this time are material and favour management. Yinwang is a profitable, revenue-generating business with a real customer base rather than a pre-revenue startup; the stake was priced at a valuation set jointly with another strategic buyer at arm's length rather than negotiated bilaterally with a founder; and it comes with governance rights.

So the SF Motors record does not reject this thesis — but it narrows it. The claim that Seres has "locked in" Huawei's innovation pipeline is not supported. A 10% minority holder does not lock in an 80% owner. What the stake genuinely buys is information, alignment, and a claim on Yinwang's profits. Whether that claim is worth RMB 11.5 billion is a question the market cannot yet answer, because Yinwang's contribution to Seres' earnings has been modest relative to the outlay and no listing has occurred to mark it. The falsifying event to watch is straightforward: if Yinwang's equity-method contribution to Seres remains immaterial through 2027 while Huawei broadens supply to rivals on equal terms, the investment will look like an access fee rather than an asset.

There is one more angle on this stake that cuts against the "locking in" language, and it is structural rather than speculative. Avatr — a Changan-backed brand competing directly with AITO for the same Chinese premium buyer — sits on the same board with the same 10% on identical terms. Whatever roadmap visibility Seres obtains, a direct competitor obtains simultaneously. Whatever priority Seres believes it has purchased, Avatr has purchased at the same price. This is almost certainly deliberate on Huawei's part: a shareholder register constructed so that no single automaker can use equity to extract preferential treatment. From Huawei's perspective that is sound governance. From a Seres shareholder's perspective it means the stake confers alignment and information, but explicitly not advantage.

The third leg of capital strategy was the Hong Kong listing. On November 5, 2025, Seres listed on the HKEX main board under 9927, selling 100.2 million H shares at HK$131.50 to raise HK$14.28 billion — around $1.8 billion, priced at the top of the range and reportedly the largest listing ever by a Chinese automaker.²⁷⁴⁴ The Hong Kong retail tranche was oversubscribed 133 times. The debut itself was flat to weak, with shares opening down and trading below issue intraday.²⁸ Proceeds were earmarked heavily toward research and development, with roughly a fifth directed to overseas market expansion and charging infrastructure, alongside plans for around 100 experience centres in Europe and the Middle East.²⁹

The gap between the retail frenzy and the flat debut is itself informative. Hong Kong's retail investors were buying a narrative — the Huawei car company, arriving at the top of a momentum run in Chinese technology equities. Institutional buyers, working from the same prospectus, priced it at issue and no higher. Within eight months the shares would trade far below the offer price, which means the institutions were closer to right and the retail multiple was paying for growth that had already decelerated to 13.7% by the time the shares were sold.¹³³ For long-term investors, the practical lesson is not about Seres specifically: an oversubscription statistic measures enthusiasm, not value, and a listing priced at the top of the range at the end of a re-rating is a seller's transaction by construction.

The listing did two things worth noting. It transformed the balance sheet — total equity rose from RMB 12.3 billion at the end of 2024 to RMB 41.9 billion a year later, against total debt of only RMB 4.6 billion and cash of RMB 87.3 billion.³⁷ And it forced disclosure. The prospectus required Seres to lay out its related-party economics with Huawei in detail for the first time, and what emerged from that document would reframe the entire investment debate within two months of listing.

↑ Back to Top

VII. Segment Breakdown, Financial Materiality & Sizing

Strip Seres down to its economic skeleton and it is close to a single-product company wearing a group structure.

The AITO franchise — the M5, M7, M8, M9, and the newer M6 — is effectively the entire business. It accounted for more than eight-tenths of group NEV sales volume in the first half of 2025 and, because AITO vehicles carry transaction prices multiples above anything else Seres sells, its share of revenue and its share of gross profit run higher still.³⁸ On any reasonable estimate, AITO represents roughly nine-tenths of revenue and the overwhelming majority of gross profit. Everything else in the portfolio is either legacy or option value.

蓝电 Landian is the mainstream NEV brand, launched in early 2023 and positioned as an accessible plug-in hybrid alternative — the E5 Plus was marketed, not unfairly, as roughly a half-priced M7.⁴² Its economic footprint is small. Its strategic function is worth understanding, though: it absorbs production capacity on non-AITO lines and serves price-sensitive domestic and export buyers without dragging AITO's premium positioning down. That is a sensible use of otherwise idle assets rather than a growth engine, and investors should size it accordingly.

The Dongfeng Sokon commercial vehicle business — the microvans and light trucks that funded everything — is now a modest, declining contributor. It supplies stable base utilization for older plants and generates cash, but it is not a source of growth and its long-run trajectory is downward as China's rural commercial fleet electrifies and consolidates. There is a quiet accounting consequence worth noting: older plants running declining volumes are the natural home for impairment charges, and a company that keeps legacy internal-combustion capacity on the books in a market shifting decisively to electric carries a standing risk of write-downs that have nothing to do with how AITO is performing.

The components and powertrain operation — range extenders, battery packs, stampings — exists primarily as internal vertical integration, with limited third-party sales. Its value is not its revenue line; it is the degree to which owning the range-extender and pack assembly keeps a slice of value inside Seres rather than flowing to suppliers. This is the one part of the technology stack where Seres has genuinely built rather than bought, and it is not trivial: the efficiency of a range extender — how much electricity the on-board generator produces per litre of fuel — is a real engineering differentiator that affects both running costs and the size of battery a vehicle needs. It is also, notably, the part of the vehicle that Huawei does not supply.

Then there is the overseas business, which management has begun to describe in terms that outrun its current economics. Seres exports commercial vehicles and passenger models to Latin America, Europe, and Southeast Asia, has launched the SERES 5 in Europe above $60,000, and has established operations in Norway, Germany, the UK, and Switzerland.²⁹ Twenty percent of Hong Kong IPO proceeds were directed toward overseas expansion and charging networks.²⁹ As of today, exports remain a small single-digit share of the business, and the honest framing is that this is plausibly material to the future investment case as a hedge against the domestic price war — but it is not yet material to earnings, and it faces an unusually specific problem.

That problem is Huawei. Outside China, Huawei's consumer brand is damaged by sanctions, its retail network is a fraction of its domestic scale, HarmonyOS has almost no installed base, and its driver-assistance stack faces regulatory regimes that have not approved it. Every mechanism that makes AITO work in Chengdu is absent in Cologne. A Seres vehicle sold in Norway is competing on the merits of the vehicle alone against Tesla, BYD, and the entire European industry, without the ecosystem that constitutes its domestic advantage.

The overseas ambition is therefore not an extension of the existing moat; it is a test of whether Seres can build a business without it. That test has barely begun, and the company's one prior attempt to sell premium vehicles outside China ended in an Indiana asset sale.

One further piece of the portfolio deserves a sentence, because it is where the company's own capital is currently pointing. Research spending in 2025 went disproportionately into a next-generation vehicle platform, a higher-efficiency range extender, artificial intelligence, and — a line item that appears in the 2026 interim disclosures without much elaboration — humanoid robotics.³⁰ The first two are core and defensible uses of shareholder money. The last is the kind of adjacency that has a poor historical conversion rate across the global auto industry, and Seres has one prior example of enthusiastically funding a technology adjacency that produced no revenue. It is small today. Investors should watch whether it stays small.

The concentration arithmetic is the section's real conclusion. A company deriving nearly all of its gross profit from one brand, sold through one partner's retail network, running one powertrain architecture, in one country, is a concentrated bet by any standard. Concentration is not automatically a flaw — it is how most great compounders are built — but it removes the buffer. When the single engine skips, as it did in the first half of 2026, there is nothing else in the portfolio large enough to absorb the shock.

Which brings us to the people who chose that structure, and how they have handled it when it faltered.

↑ Back to Top

VIII. Modern Strategy & Management Credibility Audit

Zhang Xinghai is not a Chinese tech-founder archetype. He does not do keynote theatrics, he did not build a personality cult, and the defining feature of his forty-year career is a willingness to occupy positions other executives find degrading. He made washing-machine springs. He made RMB 40,000 microvans. He agreed to build cars that another company would name, design, and sell. Chen Hong worried about SAIC's soul; Zhang appears to have concluded that a soul is a luxury good and market share is a necessity. He controls Seres through Chongqing Sokon Holdings and Yu'an Industrial, holding 50% of the holding company alongside his two brothers, and remains the company's largest shareholder.³

The second-generation question is now live. His son 张正萍 Zhang Zhengping had led the listed company as chairman through the AITO build-out and the deepening of the Huawei relationship. Then, in a sequence of changes registered in the spring of 2026, Zhang Xinghai returned as chairman of the listed entity while 尹先知 Yin Xianzhi became president, and in early June 2026 Zhang Zhengping took the chairmanship of the vehicle subsidiary, Seres Auto.³⁹⁴⁰

The company's framing is that this is a deliberate division of labour — the founder taking group strategy, capital markets, and major decisions; the son focusing on vehicles, brand, and market expansion. That may be exactly right. But an analyst should note what the disclosure did and did not contain. The changes were communicated as registered corporate facts rather than explained as strategy, and they landed in the same window as the profit warning and the share-price collapse. A founder resuming the chairmanship of a listed company weeks before it reports its first loss in three years is, at minimum, an event that warranted a fuller explanation than shareholders received. Investors should not over-read it as a crisis of confidence; they should note it as a disclosure practice that leans toward the minimum.

Now the substance of the credibility audit, which turns on how management explained the 2026 reversal.

The explanation offered was cost inflation. Zhang Xinghai pointed to memory chip prices rising from roughly RMB 20 to nearly RMB 100 per unit and lithium carbonate moving from RMB 80,000 to RMB 180,000 per tonne, adding RMB 15,000 to RMB 20,000 to the cost of every AITO vehicle.³¹³³ The company also flagged non-recurring impairment charges on tooling, inventory, and equipment made obsolete by model transitions.³⁰

Assess that on the evidence. The input-cost claim is verifiable, industry-wide, and corroborated by profit warnings from other Chinese manufacturers in the same period — it is not a manufactured excuse. But it is incomplete in a way that matters. Three other forces were at work, and management's own numbers show them. Selling expenses rose roughly 40% year-on-year in the first quarter of 2026 while revenue rose 34%, so distribution costs were growing at least as fast as the top line.³² Impairments were not genuinely non-recurring: they ran at RMB 424 million in 2023, RMB 2.18 billion in 2024, and RMB 1.58 billion in 2025 — three consecutive years of substantial write-downs from accelerating model refresh cycles.³²

A charge that appears every year at billion-yuan scale is a cost of doing business in a market with eighteen-month product cycles, not an exceptional item, and treating it as such is an accounting judgment investors should watch. And most importantly, the mix reversed: the RMB 500,000-plus M9 gave way to the RMB 300,000 M7 and the newer M6 as volume leaders.³¹

That last point is the one management has been least forthcoming about, and it is arithmetically the largest. Recall that AITO deliveries grew 10.2% in the first half of 2026 while revenue fell to RMB 57.49 billion from RMB 62.40 billion.³⁰³⁸ More units, less revenue means revenue per unit fell by roughly a sixth. The company that celebrated a RMB 400,000-plus average transaction price a year earlier had experienced material price and mix erosion, and the framing chose to lead with commodity prices.

There is a fair counterpoint on transparency: the Hong Kong prospectus process did force genuine disclosure of the Huawei economics, and management has since acknowledged the per-vehicle cost structure publicly rather than hiding behind aggregate related-party totals. That is a real improvement in disclosure discipline relative to the pre-listing period, and it was not entirely voluntary.

There is a governance dimension to this that a sceptical investor would press hard. The relationship with Huawei is, in accounting terms, a set of related-party transactions of extraordinary magnitude — RMB 22.34 billion of purchases from Yinwang in 2025 alone, from an entity in which Seres holds a 10% board-represented stake.³⁵ Related-party flows at that scale, priced by negotiation rather than by a competitive tender, are exactly the structure that minority shareholders are trained to scrutinize. There is no evidence in the public record of improper pricing, and the terms appear to be broadly consistent with those offered to other alliance partners. But the volume and the concentration mean that the single most important determinant of Seres' profitability is a commercial negotiation conducted with a counterparty that holds most of the leverage and none of the obligation to Seres' shareholders.

An activist would also note the portfolio question: a company earning a low-single-digit net margin while sitting on a RMB 11.5 billion minority stake and a large cash balance invites the argument that capital could be returned rather than accumulated. The 2025 dividend of RMB 0.8 per share, totalling roughly RMB 1.9 billion, is a start but represents under a third of earnings.¹

On alignment, the actions taken during the drawdown were substantive rather than cosmetic. The company repurchased more than RMB 587 million of stock through July 2026, executives bought approximately RMB 148 million of shares within three trading days, and the controlling shareholder committed to a further RMB 150–300 million of purchases over six months.³⁰³³ Insiders buying their own stock with their own money during a collapse is among the more meaningful signals available to outside investors. It does not prove they are right. It does establish that they are exposed to the same outcome.

The composite verdict is neither the bull's nor the bear's. This is a management team with one catastrophic capital allocation failure, one extraordinary strategic call, a pattern of explaining bad news in terms of external factors while under-emphasizing internal mix dynamics, minimal narrative disclosure around governance changes, and demonstrated personal financial commitment when the stock fell. What would move the assessment materially is not rhetoric. It is whether the next model cycle restores average transaction prices — and whether management says so plainly if it does not.

↑ Back to Top

IX. Playbook: Business & Investing Lessons

Three transferable lessons come out of this story, and each has a sharp edge that the celebratory version of the narrative tends to sand off.

Counter-positioning works precisely because incumbents cannot copy it without self-harm. Hamilton Helmer's insight is that the best strategic positions are ones where the incumbent's rational response to your move is to not respond. SAIC could not accept Huawei's Smart Selection terms in 2021 without devaluing its own brands, cannibalizing its dealer network, and telling its engineering organization that its software work was worthless.

Seres had none of those constraints. It captured the position not by being smarter but by having less to lose — and it is worth being precise about that, because the lesson for investors is that counter-positioning advantages are often born of weakness rather than insight, and they decay as the incumbents' pain from staying out exceeds their pain from joining. By 2025, SAIC had joined. So had 奇瑞 Chery, 北汽 BAIC, 江淮 JAC, and Changan.⁴¹ The position Seres captured through willingness was always going to be entered by others once the economics were demonstrated.

Borrowed distribution is astonishingly capital-efficient right up until you calculate the rent. Consider what a Huawei experience centre does that a car showroom cannot. A conventional dealership is a destination: a customer decides to shop for a car, drives to a retail park, and walks in with their guard up. A Huawei store is a place people already are, browsing phones on a Saturday afternoon in a shopping mall, and the car is simply there — sitting in a space they entered for another reason, wearing a badge they already trust with their personal data. The conversion mechanics of that are fundamentally different, and they are why the M7 relaunch could gather six-figure order volumes in weeks rather than quarters.

Building a national direct-sales network in China — the NIO or Tesla model — costs billions of renminbi in leases, fit-outs, staff, and years of ramp. Seres skipped all of it, plugging into hundreds of Huawei experience centres that already carried enormous foot traffic from consumers who came to look at phones.³⁴ The capital saving is real and it is one of the reasons the balance sheet could support this growth at all. But the disclosure that followed the Hong Kong listing quantified the rent, and it is not small: the HIMA arrangement involves a comprehensive service fee of approximately 10% of vehicle sale price, of which roughly 8 percentage points cover channel and brand marketing and roughly 2 points cover technology licensing.³⁴ Add hardware procurement and the total flow is far larger.

The lesson generalizes well beyond automobiles: platform distribution converts a capital expense into a permanent revenue-share, and revenue-shares scale with your success while capital costs amortize away. Which structure is superior depends entirely on how long you intend to be in business.

Cutting losses is a skill, but it is a lesser skill than not making the loss. The Seres playbook is often praised for the discipline of abandoning the US programme and redeploying into the Huawei partnership. The redeployment was excellent. The abandonment was late and forced. The honest version of this lesson is that the company's near-death experience in America was what made it willing to accept terms that its better-capitalized peers refused — the failure created the opportunity. That is a lesson about survival and adaptation, not about foresight, and investors should be sceptical of any framing that presents this management as having planned the sequence.

The common thread is that Seres' greatest strategic asset has been an accurate assessment of its own weakness. That is rarer among executives than it sounds, and it is genuinely valuable. It is also, structurally, not a moat.

There is a fourth lesson embedded in the first three that is more useful to investors than any of them individually: when a company's economics depend on a contract rather than on an asset, the terms of that contract are the investment thesis, and you should refuse to underwrite the business until you can see them. For four years, public investors in Seres were valuing a partnership whose commercial terms had never been disclosed in detail. The revaluation that followed the Hong Kong prospectus and the subsequent reporting on per-vehicle payments was not caused by the terms getting worse — the terms had always been what they were. It was caused by the terms becoming visible at the same moment that growth slowed enough for the arithmetic to bite. Opacity is not a risk that shows up in a ratio. It shows up all at once.

↑ Back to Top

X. Strategic Analysis: 7 Powers, 5 Forces, Bear vs. Bull & Risk Radar

Myth versus reality: three consensus claims, tested.

Myth one: Seres is a Huawei subsidiary in all but name. Reality: Huawei holds no equity in Seres. The relationship runs entirely through commercial contracts — component supply, technology licensing, and a channel service fee — and the AITO trademarks now sit on Seres' balance sheet.²² This matters in both directions. It means Seres genuinely is an independent listed company whose shareholders capture the residual. It also means there is no ownership tie compelling Huawei to prioritize Seres over its four other alliance partners.

Myth two: the 2026 loss was a one-off commodity shock. Reality: input costs were a genuine and quantified driver, but the same period saw distribution costs growing faster than revenue, a third consecutive year of billion-yuan-scale impairments, and a sharp fall in revenue per vehicle.³² Only one of those four is cyclical.

Myth three: Seres is a low-margin assembler with no real business. Reality: this is the bear case pushed past the evidence. A pure assembler does not achieve a 28.8% NEV gross margin, does not lead the RMB 500,000-plus segment for two consecutive years, and does not generate RMB 28.91 billion of operating cash flow in a single year.¹ Seres captures less of the value chain than its brand position suggests, but it captures materially more than a contract manufacturer would.

Hamilton Helmer's 7 Powers, applied honestly.

Counter-Positioning was Seres' strongest power and it is now substantially spent. The evidence is unambiguous: AITO's share of HIMA sales fell from roughly 87% in 2024 to approximately 60% by June 2026 as Luxeed, Stelato, Maextro, and Shangjie filled the same showrooms.³³ Seres was first through the door and no longer has the room to itself.

Scale Economies are building and are the most credible durable advantage. Amortizing plant capital and shared platform development across roughly 470,000 annual NEV units gives Seres a genuine unit-cost position relative to smaller premium challengers.² The caveat is that scale in autos is only an advantage against smaller rivals, and BYD, Geely, and Changan all operate at multiples of Seres' volume.

Process Power is moderate and real. The M7 ramp to 700 units a day within weeks of a demand surge, and the M9's ability to deliver over 10,000 units within three weeks of a new-generation launch, reflect manufacturing execution that most startups cannot match.¹⁴³⁰ This is the direct inheritance from the microvan years.

Cornered Resource is the weak point, and it is weak in a specific way. Seres does not own the resource that makes its product desirable — Huawei does. The Yinwang stake is an attempt to partially corner it, and as argued above, a 10% minority position alongside an identical stake held by a direct competitor does not corner anything. It buys a seat, not control.

Switching Costs operate at the ecosystem layer rather than the vehicle layer, and this distinction is the crux of the bull case. A consumer deeply embedded in HarmonyOS — phone, tablet, watch, car — faces genuine friction leaving. But that friction binds them to Huawei, not to Seres. Within the ecosystem, moving from an AITO M7 to a Luxeed R7 costs the customer nothing. Seres captures the benefit of Huawei's switching costs only for as long as it offers the most compelling vehicle inside the alliance.

Branding is emerging. AITO has established real price authority in the RMB 400,000–600,000 band, something no Chinese brand had achieved, and now owns its trademarks. Whether that brand equity is transferable away from Huawei is untested and, on the evidence of the SF5, doubtful.

Network Economies do not apply to Seres directly. They apply to HarmonyOS, and Huawei owns them.

Porter's Five Forces.

Supplier power is the dominant force in this business, and it is extreme. Huawei is not a supplier in the ordinary sense; it is simultaneously the supplier of critical components, the licensor of the software, the owner of the distribution channel, and the marketer of the product. Seres paid Huawei's Yinwang unit RMB 22.34 billion for goods in 2025, and total flows to the Huawei system exceeded RMB 30 billion — more than a fifth of automotive revenue and several times the group's RMB 5.96 billion net profit.³⁵ Cumulative procurement payments from 2022 through the first half of 2025 exceeded RMB 75 billion.³⁴ On a per-vehicle basis the figures reported from the Hong Kong prospectus disclosures range from roughly RMB 92,000 to RMB 136,000 depending on what is included and which period is used — in the higher framing, over 23% of the selling price.³³³⁴

宁德时代 CATL supplies the batteries and holds its own considerable leverage. There is no plausible reading of these numbers in which Seres holds the balance of power in its own supply chain.

The one qualification worth attaching is that supplier power in this relationship is bounded by mutual interest rather than by contract. Huawei's automotive unit needs volume to amortize its own development spending, and AITO has been by a wide margin its largest source of it. That is a real constraint on how hard Huawei can squeeze, and it is why the terms have reportedly stayed broadly stable across partners rather than being renegotiated upward as AITO grew. But a constraint founded on the partner's self-interest is a weaker protection than a contractual one, and it weakens further with every additional alliance brand that reaches scale.

Buyer power is high and rising. Chinese premium EV buyers face an extraordinary array of choice, product cycles of eighteen months, and a normalized price war. The M7's original relaunch success was partly bought with a price cut; the 2026 mix collapse shows buyers trading down within the range when a cheaper option appears.

Substitutes are abundant. Li Auto occupies the same range-extended family-SUV territory. 蔚来 NIO competes on service and battery swap. 比亚迪 BYD attacks from below with Denza and from above with Yangwang. German premium ICE and EV products retain residual brand appeal. And the intra-HIMA substitutes are the most dangerous of all, because they sit on the same shop floor.

New entrants face high capital barriers but the barriers do not stop the entrants that matter. 小米汽车 Xiaomi Auto demonstrated that a consumer-technology company with capital, brand, and an existing user base can enter and scale rapidly. The barrier protects Seres from startups, not from platforms.

Rivalry is as intense as any major industry on earth. This is the base condition, not a risk factor.

A word on the competitive set, because the comparison clarifies the position. Li Auto pioneered the range-extended premium family SUV in China and built its own direct retail network — higher capital intensity, but it owns the customer relationship and the data that comes with it. NIO built battery-swap infrastructure and an unusually loyal owner community, at enormous and sustained cash cost. XPeng invested early and heavily in its own autonomous-driving software. BYD vertically integrated batteries, semiconductors, and vehicles at a scale that gives it the lowest cost position in the industry and lets it attack any price band it chooses. Xiaomi arrived with a consumer brand, an existing device ecosystem, and enough capital to absorb years of losses.

Each of these companies made a different bet about which layer of the automobile would be scarce. Seres is the only one of the group that bet the scarce layer would be somebody else's, and that the right move was to buy access rather than build it. That bet produced the fastest revenue growth of any of them and, so far, among the thinnest net margins. Both of those facts follow from the same decision.

The peer comparison also sharpens the falsification question. Li Auto's owned network means that if its product falters, it still controls the storefront and can relaunch; Seres' storefront is leased and shared. BYD's vertical integration means input-cost shocks compress its margins less than they compress a company that buys batteries and intelligent hardware from others — which is precisely the mechanism that produced the 2026 loss. Neither observation makes Seres' strategy wrong. Both make it more fragile to specific, identifiable shocks, and both explain why the same commodity cycle that dented the industry pushed Seres from profit to loss while several peers merely saw margins narrow.

The bull case, stated at its strongest.

Seres has done something no other Chinese manufacturer has: established sustained pricing authority in the RMB 400,000-plus segment, displacing German incumbents in their last profitable stronghold, and it delivered over 420,000 AITO vehicles in 2025 with a 28.8% NEV gross margin.¹ It reached the second-highest cumulative-delivery milestone in Chinese premium NEVs — one million AITO vehicles in 46 months.³⁰

It is now capitalized as few Chinese automakers are, with RMB 87.3 billion of cash against RMB 4.6 billion of debt after the Hong Kong raise, giving it the ability to fund R&D through a downturn while weaker rivals consolidate.³⁷ R&D spending rose 77.4% to RMB 12.51 billion in 2025 with research headcount up 45.4%, which is at least directionally consistent with an attempt to build proprietary capability rather than remain a pure assembler.¹ The Yinwang stake gives it economics from Huawei technology sold to competitors. And the input-cost shock of 2026 is cyclical: lithium and memory prices mean-revert, and when they do, roughly RMB 15,000–20,000 per vehicle of cost returns to the margin line.³¹

The bull's strongest structural point is subtler than any of those. Huawei's alliance succeeds or fails on the quality of the vehicles underneath the software, and Seres has demonstrated — through the M7 ramp, the M9's segment leadership, and a manufacturing organization inherited from the microvan era — that it is the most operationally capable builder in the group. Huawei has an interest in its highest-volume, highest-quality partner remaining healthy. Dependence runs in both directions, even if it runs far more strongly in one.

The bear case, stated at its strongest.

The economics of the partnership mean Seres is structurally a low-net-margin manufacturer regardless of how premium its products look — 3.6% net margin in its best year.³³ Its share of the alliance it pioneered has fallen from roughly 87% to 60% in eighteen months as Huawei added four competing partners, and Huawei has every incentive to keep adding them.³³⁴¹ The 2026 loss revealed that the business has almost no operating buffer: a commodity shock and an adverse mix shift, together, were sufficient to erase profitability entirely despite volume growth.³⁰³¹ Overseas expansion is being funded before it is proven, into markets where the partner that makes the domestic model work is a liability rather than an asset.

And there is a balance-sheet subtlety that deserves an activist's attention. That RMB 87.3 billion cash pile sits against RMB 79.6 billion of accounts payable on RMB 165 billion of revenue — roughly six months of cost of goods sold in supplier credit.³⁷ Much of the cash is supplier float, not owner's capital. The mechanism is benign while volumes grow and turns hostile when they stall: first-half 2026 operating cash flow was negative RMB 20.95 billion as payables unwound faster than receipts came in.³² Seres is not financially fragile — net cash of that magnitude is a genuine cushion — but investors reading the headline cash balance as freely deployable firepower are misreading a working-capital position as a war chest.

Weighing it. The historical record does not reject the claim that Seres built a real premium franchise — the M9's segment leadership over two consecutive years, sustained at prices above RMB 500,000, is not something a contract assembler achieves.¹⁴⁶ But the record substantially narrows the claim that Seres owns a defensible position. The moat belongs to Huawei; Seres rents a share of it, at a disclosed rate, on a lease that Huawei has already sub-let to four competitors. The most defensible version of the bull thesis is therefore not "Seres has a moat" but "Seres is the most operationally capable tenant in the best building in Chinese autos, and tenancy in a great building can be worth a great deal." That is a real thesis. It is simply a different one, and it should be valued differently.

Three KPIs that settle the argument.

AITO's share of HIMA deliveries, together with blended average transaction price. This single pairing captures both competitive threats at once — intra-alliance cannibalization and mix-down. If AITO's alliance share stabilizes near 60% and average transaction price recovers toward RMB 400,000, the premium franchise is intact. If share keeps sliding while price falls, Seres is becoming the alliance's volume brand, which is a fundamentally lower-value position.
Automotive gross margin against operating margin, tracked together. Gross margin alone is misleading given where the Huawei channel fee sits. The spread between the two is the cleanest available measure of how much of the value Seres actually keeps.
Yinwang's equity-method contribution to Seres earnings. This is the direct test of whether RMB 11.5 billion bought an asset or an access fee. It is a number readers can track in each interim and annual report without calculating anything.

Risk radar. Partner concentration remains the dominant structural risk and requires no elaboration. Beyond it: HIMA's total deliveries fell 5.52% year-on-year in August 2026, a third consecutive monthly decline, meaning the alliance itself — not just AITO's share of it — has stopped growing.³⁶

Chip supply is a live geopolitical exposure, since restrictions on advanced processors for autonomous driving would hit Huawei first and Seres immediately after. Input costs are cyclical but the company has demonstrated no ability to pass them through. Product-transition vulnerability is now a proven pattern rather than a hypothetical, with three straight years of billion-yuan-scale impairments tied to model refreshes.³² And the governance profile — a controlling family, minimal narrative disclosure on leadership changes, and enormous related-party flows to a single counterparty — is one where minority shareholders are structurally dependent on the alignment of interests continuing to hold.

↑ Back to Top

XI. Epilogue & Key Takeaways

There is a photograph, of the kind that circulates on Chinese business social media, of Zhang Xinghai standing in a Huawei store beside a car with his company's chassis and another company's soul. It is easy to read it as capitulation. It is more accurate to read it as arithmetic.

Seres is the clearest case study available of what happens when a manufacturer accepts that the most valuable part of its product has migrated to software and brand, and decides to rent that layer rather than lose to it. The rent is high — roughly a tenth of the sale price in service fees, plus the components, plus the marketing, adding to a flow that in 2025 exceeded a fifth of automotive revenue.³⁵ The alternative, which the company tested exhaustively and expensively in Santa Clara and Mishawaka, was worse. Between paying rent and going bankrupt, paying rent is the correct decision. That is a genuine strategic insight and it is why the company exists today with RMB 165 billion of revenue instead of as a footnote in the history of Chinese microvans.

But the first half of 2026 clarified the boundaries of the achievement. A business can grow units and shrink profits. A pioneer's advantage inside an alliance can be diluted by the alliance's own success. A cash mountain can be someone else's money. And a moat that belongs to your partner is, from a shareholder's perspective, a moat you are exposed to rather than protected by.

There is a broader lesson here for anyone investing in manufacturing businesses in an era when software defines the product. The automobile is only the most visible case; the same migration is underway in medical devices, industrial equipment, agricultural machinery, and consumer appliances. In every one of those industries, some firms will build the intelligent layer, some will be commoditized by it, and some will do what Seres did — accept a subordinate position in the value chain in exchange for volume, and try to climb back up over time. The Seres experiment is the most advanced live test of whether that third path leads anywhere. Its 2025 accounts say the path is survivable and can produce real profits. Its 2026 accounts say the path leaves almost no margin for error.

It is also worth naming what would constitute genuine escape velocity, as distinct from a good quarter. Escape velocity for Seres does not look like higher volumes; it looks like a widening gap between gross margin and operating margin closing in the company's favour, which would mean the rent is falling as a share of the value created. It looks like a vehicle that sells well because of something Seres built — a platform, a range extender, a manufacturing cost position — rather than something Huawei installed. And it looks like a second geography where the company earns a premium price without the ecosystem. None of those three has happened yet. All three are things the company says it is working toward, and management's stated ambition of reaching a second million cumulative deliveries within two years is a volume target rather than a value-capture target.¹ The distinction matters more than the number.

What to watch from here is concrete. Whether the next generation of AITO products restores average transaction prices, or whether the M9's dominance at the top of the market proves to have been a moment rather than a position. Whether Huawei's rollout of increasingly autonomous driving capability continues to reach Seres first, or arrives simultaneously across all five HIMA partners — the latter being the strategically meaningful outcome. Whether the overseas build-out, funded with Hong Kong capital, converts into vehicles delivered at profitable prices in markets where the Huawei ecosystem does not exist. And whether Yinwang, in time, delivers earnings rather than merely optionality.

Seres surrendered the soul and gained the scale. The unresolved question — the one that a decade from now will determine whether this was one of the great strategic pivots in industrial history or an unusually well-executed contract manufacturing arrangement — is whether a company can rent an identity long enough to eventually afford one of its own.

Seres Group Co.,Ltd.

Table of Contents

Seres Group: The Huawei Handshake and the Twenty-Billion-Dollar Auto Turnaround

I. Introduction & Episode Roadmap

II. The Chongqing Roots: Springs, Shock Absorbers & Microvans

III. The Silicon Valley Detour & The $500M US EV Gamble

IV. The Faustian Bargain: Handing the Keys to Huawei

V. The `问界 AITO` Inflection & Financial Turnaround

VI. Capital Allocation & Ecosystem Locking: AITO Trademarks & Shenzhen Yinwang

VII. Segment Breakdown, Financial Materiality & Sizing

VIII. Modern Strategy & Management Credibility Audit

IX. Playbook: Business & Investing Lessons

X. Strategic Analysis: 7 Powers, 5 Forces, Bear vs. Bull & Risk Radar

XI. Epilogue & Key Takeaways

References

Seres Group Co.,Ltd.

Table of Contents

Seres Group: The Huawei Handshake and the Twenty-Billion-Dollar Auto Turnaround

I. Introduction & Episode Roadmap

II. The Chongqing Roots: Springs, Shock Absorbers & Microvans

III. The Silicon Valley Detour & The $500M US EV Gamble

IV. The Faustian Bargain: Handing the Keys to Huawei

V. The 问界 AITO Inflection & Financial Turnaround

VI. Capital Allocation & Ecosystem Locking: AITO Trademarks & Shenzhen Yinwang

VII. Segment Breakdown, Financial Materiality & Sizing

VIII. Modern Strategy & Management Credibility Audit

IX. Playbook: Business & Investing Lessons

X. Strategic Analysis: 7 Powers, 5 Forces, Bear vs. Bull & Risk Radar

XI. Epilogue & Key Takeaways

References

V. The `问界 AITO` Inflection & Financial Turnaround