LONGi Green Energy: The Photovoltaic Titan and the Solar Supercycle

I. Introduction & Episode Roadmap

In the spring of 1986, two teenagers stood in a courtyard at 兰州大学 Lanzhou University, in one of the poorest and driest corners of northwest China, at a ceremony held in front of a statue. The statue was of 江隆基 Jiang Longji, a former president of the university who had died two decades earlier during the Cultural Revolution, and whose reputation among Lanzhou students rested on a single stubborn quality: an insistence that scientific truth did not bend to political fashion. The two students, 李振国 Li Zhenguo and 钟宝申 Zhong Baoshen, were physics majors. Fourteen years later, when they founded a company together, they named it after him.¹

That is the origin myth, and like all origin myths it is retold by the company because it is useful. But it is worth holding onto, because the whole LONGi story is a series of moments where the company decided that the physics was right and the consensus was wrong — and then bet the balance sheet on it. Sometimes spectacularly correctly. Sometimes, as the last three years have shown, at ruinous cost.

By 2022, that approach had built something genuinely extraordinary. 隆基绿能科技股份有限公司 LONGi Green Energy Technology Co., Ltd. — known at home simply as 隆基绿能 LONGi Green Energy, and listed in Shanghai as 601012 — booked revenue of RMB 129 billion and net profit of RMB 14.81 billion, up 63% year on year, on shipments of 85 GW of monocrystalline wafers and nearly 47 GW of modules.² It was the largest solar manufacturer on earth by almost any measure you cared to pick. At the end of 2021 the market valued it at roughly RMB 465 billion.

By September 2026, the same company carries a market capitalisation of around RMB 89 billion. It has just posted its eleventh consecutive quarterly loss. Cumulative losses since the fourth quarter of 2023 run to roughly RMB 19 billion. Roughly a quarter of the book equity that existed at the end of 2023 has been erased. And the founder who ran the company for twenty-five years stepped down as general manager and legal representative in May 2025 to focus on research.[^3]

The central question this article tries to answer is narrow and uncomfortable. LONGi's entire investment case now rests on a single proposition: that its bet on 背接触 BC (Back Contact) cell architecture — made after it lost the race to TOPCon — is the same kind of contrarian technical call that made it a giant in the first place, and will make it a giant again. Management says so explicitly and repeatedly. The company has built roughly 60 GW of BC capacity to prove it.

That proposition deserves to be tested, not repeated. And the test is not whether BC cells are better physics — they probably are — but whether a differentiated technology can earn a durable return in an industry where the previous differentiated technology, LONGi's own monocrystalline wafer, was competed away into a commodity within six years of winning.

Here is the road we will travel:

The Lanzhou pact, and why a levelised-cost-of-electricity argument justified twenty years of contrarianism.
The twin process breakthroughs — repeated Czochralski crystal pulling and 金刚线切割 diamond wire sawing — that made monocrystalline silicon cheap and killed the polysilicon incumbents. And the awkward fact that LONGi destroyed its own moat in the process.
The 2014 acquisition of 乐叶光伏 LERRI Solar, a small, struggling Zhejiang module maker, and how it turned a component supplier into the world's biggest solar brand.
The overcapacity catastrophe of 2023–2026, the unit economics of an industry selling below cash cost, and how LONGi's losses compare with 通威股份 Tongwei, 晶科能源 JinkoSolar, 晶澳科技 JA Solar, 天合光能 Trina Solar, TCL中环 TCL Zhonghuan and 爱旭股份 Aiko Solar.
The BC-versus-TOPCon war, and an honest audit of management's track record on promises made and kept.
The small optionality bucket — 隆基氢能 LONGi Hydrogen, and now energy storage — sized to what it actually earns.
A 7 Powers and Porter analysis, the bull and bear cases, and the two or three numbers that will actually settle the argument.

Let us start where the physics started: in a lab, with a crucible of molten silicon, and a decision nobody else wanted to make.

↑ Back to Top

II. The Lanzhou Pact & Monocrystalline Conviction (2000–2011)

Li Zhenguo graduated from Lanzhou University's physics department in 1990, having majored in semiconductor materials.¹ This matters more than the usual founder-biography detail, because the defining trait of LONGi's first decade was that it was run by someone who thought about solar panels the way a semiconductor engineer thinks about wafers — as a crystal-growth problem — rather than the way an energy entrepreneur thinks about them, as a subsidy-arbitrage problem.

The company was incorporated on 14 February 2000 as Xi'an Xinmeng Electronic Technology, a modest silicon-materials outfit in Shaanxi province.³ Li was joined over the following years by Zhong Baoshen and 李春安 Li Chunan, both Lanzhou classmates. The name change to LONGi came later, but the intent was set at the founding: this would be a company built around materials science, staffed by people who had studied together, and named for a man remembered for refusing to say convenient things.

The consensus they chose to ignore

To understand how contrarian LONGi's early positioning was, you have to understand the economics of solar manufacturing in the 2000s. A solar panel starts as polysilicon — purified silicon, essentially very expensive sand. That polysilicon has to be turned into a solid block, then sliced into wafers, then processed into cells, then assembled into modules.

There were two ways to make the block. You could pour molten silicon into a big square mould and let it cool, producing a multicrystalline (or "poly") ingot — a block made of many crystal grains fused together, cheap and fast, but with grain boundaries that trap electrons and drag down efficiency. Or you could use the Czochralski process: dip a tiny seed crystal into molten silicon and slowly pull it upward while rotating, coaxing the melt to solidify as a single continuous crystal lattice. That produces monocrystalline silicon — structurally perfect, more efficient, and dramatically more expensive, because pulling a crystal is slow, energy-hungry, and yields a cylinder that has to be squared off, wasting material.

The industry consensus was that poly won. It was roughly four-fifths of global PV production. Casting was simpler, capex per gigawatt lower, and in an industry where module buyers priced almost entirely on RMB-per-watt, the efficiency premium of mono did not pay for its cost premium.

The LCOE argument

Li's counter-argument was a first-principles one that today reads as obvious and at the time read as academic. Solar power is not sold by the panel; it is sold by the kilowatt-hour. The right metric is levelised cost of electricity — total lifetime project cost divided by total lifetime energy produced.

And here is the thing about a solar farm: an enormous share of its cost has nothing to do with the panels. Land, racking, foundations, cabling, inverters, labour, grid connection — these scale with area, not with watts. So if you can put more watts into the same square metre by raising cell efficiency, you spread every one of those fixed area-based costs across more electricity. A one-percentage-point efficiency gain compounds down the entire stack.

This is why efficiency mattered more than the industry believed, and why LONGi concluded that monocrystalline would eventually win if and only if someone could crush its cost penalty. The company therefore did something unusual: it refused to hedge. It made only monocrystalline wafers, accepted years of low market share, and put its engineering effort into a single question — how do you make a perfect crystal cheaply?

The analytical point for investors is that this was not a bet on a product. It was a bet on a cost curve — a wager that a process gap could be closed by sustained engineering, and that whoever closed it would own a step-change in the market. That distinction matters enormously when we get to BC, because the BC bet is a different animal: it is a bet on efficiency ceiling, with an acknowledged cost penalty that management is still working to close.

Surviving the first great bust

Conviction alone would not have been enough. Between 2008 and 2012 the solar industry went through a near-extinction event. The global financial crisis choked project finance. Then Germany and Italy — which had underwritten the first solar boom with generous feed-in tariffs — cut subsidies sharply. Demand growth stalled while capacity, built on debt during the boom, kept arriving. Suntech, once the world's largest module maker, collapsed. LDK Solar, which had bet the company on polysilicon and ingot capacity, collapsed. Germany's Q-Cells went into insolvency.

LONGi survived, and the reason is less romantic than the founding myth: it had not levered up. It had not built speculative polysilicon capacity into a boom. It had stayed small enough and narrow enough that when prices collapsed, its fixed-cost base did not kill it. Focus is usually described as a strategic virtue; in a capital-intensive commodity bust it is mainly a balance-sheet virtue.

It is worth flagging, because it becomes a recurring theme: LONGi has twice been praised for conservative expansion during an industry mania. It earned that praise in 2011. As we will see, it did not entirely earn it in 2022.

By 2012 the survivors were consolidating, mono was still a niche, and LONGi had a technology thesis and no capital. That was about to change.

↑ Back to Top

III. The Twin Technological Breakthroughs: RCZ & Diamond Wire (2012–2017)

On 11 April 2012, LONGi listed on the main board of the 上海证券交易所 Shanghai Stock Exchange under the code 601012 — into one of the worst solar markets in living memory.³ The timing looks reckless and was in fact the point: the company needed capital not to expand volume but to fund a manufacturing R&D programme whose payoff was several years out. Listing at the bottom of a cycle meant a low valuation, but it also meant the proceeds bought equipment at distressed prices and hired engineers other firms were shedding.

What followed over the next five years was one of the more consequential process-engineering campaigns in modern industrial history, and it happened in two acts.

Act one: teaching the crucible to work overtime

The Czochralski process has a structural inefficiency that had irritated crystal growers for decades. You load the quartz crucible with polysilicon, melt it, pull one ingot, then let everything cool down, break out the crucible, and start again. The heating and cooling cycle wastes enormous amounts of electricity, and the crucible — an expensive consumable — gets used once.

Repeated Czochralski, or RCZ, solves this by recharging fresh polysilicon into the still-molten crucible and pulling another ingot without ever cooling down. Think of it as the difference between heating an oven for one loaf of bread and baking five loaves back-to-back on a single heat-up.

LONGi pushed this harder than anyone. Li Zhenguo has claimed the company achieved as many as five ingot pulls per crucible against an industry norm of around three, and described "hundreds" of iterative improvements to the ingot process.⁴ Each additional pull spreads the crucible cost and the energy of the initial melt across more kilograms of output. In a business where electricity is one of the largest single line items, this was not a marginal gain.

Act two: cutting silicon with diamonds

The second breakthrough was more visible and, commercially, far more violent.

Once you have an ingot, you have to slice it into wafers roughly the thickness of a sheet of paper. The legacy method used a steel wire dragged through a slurry of silicon carbide grit and glycol — essentially, sawing with liquid sandpaper. It was slow, filthy, and wasteful. Every cut turned a band of silicon into unrecoverable dust, a loss called kerf. Slurry sawing had thick kerf and slow line speeds, and the slurry had to be handled, filtered and disposed of.

The alternative — a steel wire with industrial diamond particles bonded directly to it — cut two to three times faster, produced thinner kerf, and eliminated the slurry entirely. It had been developed in Switzerland and Japan and had one fatal commercial problem: imported diamond wire cost roughly RMB 500 per kilometre, which made the arithmetic impossible at solar-industry margins.

LONGi's response was not to invent the technology. It was to industrialise it. The company worked directly with domestic Chinese wire producers and slicing-machine builders to localise the entire supply chain, and by pulling volume through them drove the wire price below RMB 50 per kilometre — a roughly ninety percent reduction.⁵ LONGi describes itself as the first company to widely deploy diamond wire at scale, and the claim is broadly accepted in the industry.⁴

The tipping point, and the moat LONGi gave away

Combine the two. RCZ cut the cost of growing a perfect crystal; diamond wire cut the cost of turning it into wafers and cut the silicon wasted per wafer. Wafers got thinner. Kerf shrank. Grams of polysilicon per watt fell hard. And mono's cost premium over poly — the single reason the industry had rejected it — evaporated.

The market moved with startling speed. Monocrystalline was under 30% of sales in early 2016; by mid-2018 mono and multi were roughly at parity; monocrystalline cells were about 35% of the market in 2015 and roughly 95% by 2021.⁴⁶ The polysilicon-wafer incumbents who had built casting capacity — and who could not convert cast furnaces into crystal pullers — were structurally stranded. Many did not survive.

LONGi's financials tell the story cleanly. Revenue went from RMB 3.7 billion in 2014 to RMB 16.4 billion in 2017 and RMB 32.9 billion in 2019, with net profit rising from under RMB 300 million to RMB 5.3 billion over the same span.

But here is the part that usually gets left out of the hagiography, and it is the single most important lesson in this entire article. LONGi's diamond-wire achievement was to build a domestic Chinese supply chain for slicing equipment and consumables. Once that supply chain existed, it sold to everybody. LONGi's own materials note that the technology saved the whole industry on the order of RMB 30 billion a year in production costs.⁴ The company converted a proprietary advantage into an industry-wide utility — and captured only a few years of excess return before every competitor bought the same wire from the same suppliers.

That is the structural condition of solar manufacturing, and it has not changed. Process innovations diffuse. Equipment vendors are the permanent beneficiaries. A manufacturer's edge is measured in quarters of lead time, not decades of defensibility.

Which raises the obvious question about what LONGi did next: if you cannot defend a process, can you at least defend a market position?

↑ Back to Top

IV. Downstream Integration: The LERRI Acquisition & Module Dominance (2014–2020)

By 2014 LONGi had a very specific and very frustrating problem. It could make monocrystalline wafers better and cheaper than anyone. It could not get anyone to buy them.

The reason was not technical. It was balance-sheet politics. The large module manufacturers of the day — JinkoSolar, Trina Solar, JA Solar — had built enormous production lines optimised for polysilicon wafers. Switching to mono meant retooling cell lines and, more painfully, admitting that a large slice of recently-installed equipment was obsolete. Nobody writes off their own factory to help a supplier sell a better input. Incumbents defend their asset base; that is what incumbency means.

So LONGi faced the classic upstream trap: a superior component, expanding capacity, and a distribution channel controlled by people with every incentive to say no.

Buying the smallest possible key

The solution, executed in 2014, was to stop asking permission. LONGi merged with Zhejiang LERRI Solar Technology, a small cell-and-module maker founded in 2007 with registered capital of RMB 350 million, and extended its business into cells and modules.³⁷ The purchase consideration was not disclosed.

As M&A, this is worth studying precisely because it was not a transformative megadeal. LERRI was a struggling minor player. What LONGi bought was not capacity or brand — it was permission to build a demand channel for its own wafers without negotiating with its competitors. It then injected capital, rebuilt the operation around high-efficiency PERC cells fed by in-house mono wafers, and went directly to the people who actually cared about the LCOE argument: utility-scale project developers, who own the land and the racking and therefore feel every square metre of balance-of-system cost.

The pitch to those developers was not "our panels are better." It was "our panels lower your cost per kilowatt-hour, here is the model, run it yourself." That is a fundamentally different sale, and it worked because the customer's economics — not the manufacturer's marketing — did the arguing.

The business was later rebranded LONGi Solar, and in February 2017 the parent renamed itself LONGi Green Energy Technology to reflect that it was no longer a silicon-materials company.⁶ By 2020 LONGi was the world's largest module supplier, a position it held on an annual basis for several consecutive years.²

What this says about the company's strategy is more interesting than the outcome. Forward integration here was not empire-building; it was a targeted response to a specific channel blockage, executed at the smallest scale that would work. That is a genuinely good piece of corporate strategy, and it is fair to say so. It is also fair to note the consequence: LONGi stopped being a focused wafer specialist and became a vertically integrated manufacturer with fixed costs at every stage of the chain — which is exactly the structure that amplifies losses when prices collapse. The 2014 decision that made LONGi enormous is part of why 2024 hurt as much as it did.

Building an offshore hedge

The second strand of this period was geographic. LONGi opened its first overseas production base in Malaysia in 2016.³ In February 2020 it agreed to acquire Vietnam-based cell and module OEM Vina Solar for RMB 1.78 billion, around US$253 million⁸ — a business with roughly 3 GW of p-type PERC cell capacity and 7 GW of module assembly.⁹

The logic was trade defence. US tariffs on Chinese solar — Section 201 safeguards, Section 301 duties, and a long-running anti-dumping and countervailing duty regime — made direct Chinese-origin shipments uneconomic. Southeast Asian manufacturing let LONGi serve the highest-margin market in the world.

That hedge worked for several years and then stopped working, in two distinct ways.

First, forced-labour enforcement. Under the Uyghur Forced Labor Prevention Act, US Customs and Border Protection detained solar imports from LONGi among others for inspection; the company's own chairman has described the impact as a major impairment.¹⁰ Southeast Asian assembly does not help if the polysilicon upstream is presumed to originate in Xinjiang and the importer must prove otherwise.

Second, the tariff wall simply moved. On 21 April 2025 the US Department of Commerce issued final anti-dumping and countervailing determinations covering Vietnam, Malaysia, Thailand and Cambodia, with combined rates ranging from double digits into the hundreds and, for Cambodia, into the thousands of percent.¹¹ The Southeast Asian hedge had a shelf life, and the shelf life was about five years.

LONGi's more durable answer was to manufacture inside the United States. In March 2023 it formed Illuminate USA, a joint venture with developer Invenergy, to build a 5 GW module assembly plant at Pataskala, Ohio; Invenergy committed more than US$600 million and serves as anchor customer.¹² It is a sound structure — a US partner, a captive offtaker, domestic content — but it is assembly, not cells, and it does not resolve the upstream provenance question.

For investors, the durable lesson of 2014–2020 is that LONGi solved a distribution problem brilliantly and a geopolitical problem only temporarily. Trade barriers are not a moat you can buy your way around with one acquisition; they are a moving target that reprices your asset base every few years. And by 2021, LONGi was about to discover that a far larger threat was building — not in Washington, but three hundred kilometres from its own factories.

↑ Back to Top

V. Financial Structure, Segment Breakdown, & The Overcapacity Crisis (2021–2025)

The peak arrived, as peaks do, looking like a plateau.

In 2022 LONGi reported revenue of RMB 129 billion, up 60%, and net profit attributable to shareholders of RMB 14.81 billion, up 63%. Operating cash flow nearly doubled to RMB 24.37 billion. The company shipped 85.06 GW of monocrystalline wafers — first in the world for the ninth consecutive year — and 46.76 GW of modules, first for the third consecutive year.² Every operating metric said the flywheel was accelerating.

Two years later the same company lost RMB 8.62 billion.

How the segments actually work

Before the crisis, understand the machine. LONGi's revenue splits broadly into two engines and a rounding error.

Cells and modules — the finished panels sold to developers, installers and distributors — are the volume and brand business, and the largest share of revenue. In the first half of 2026 the company shipped 29.93 GW of modules, with more than 65% of module revenue coming from outside China.¹³

Monocrystalline wafers are the legacy engine and, historically, the highest-margin part of the business. LONGi shipped 111.56 GW of wafers in 2025.¹⁴ Crucially, most of that is consumed internally — in the first half of 2026, 48.91 GW of wafer shipments included only 18.98 GW sold externally.¹³ The merchant wafer business is a price-taker; the internal transfer is a cost input to modules. When wafer prices fall below cash cost, the segment is a loss centre on external sales and a cost benefit internally — which is why headline segment margins in this industry can mislead.

LONGi Hydrogen and accessories account for a low single-digit percentage of group revenue. We size it properly in Section VII rather than inflating it here.

The anatomy of an overcapacity catastrophe

What happened between 2022 and 2025 is the clearest recent example of a phenomenon every industrial investor should internalise: explosive end-demand growth is not the same as pricing power.

China's 双碳 dual carbon goals — peak carbon before 2030, neutrality before 2060 — turned solar manufacturing into a national priority. Provincial and municipal governments competed to attract PV factories with cheap land, subsidised power and local financing. Capital flooded in from every direction: incumbents, adjacent chemical companies, and entirely new entrants. Because a module line is comparatively cheap and fast to build, supply responded far faster than demand.

The result was a global manufacturing base far larger than global demand could absorb. Prices did what prices do. Module quotes that had run around RMB 1.8–2.0 per watt in 2022 collapsed, bottoming near RMB 0.6 per watt — well below the roughly RMB 1.0 per watt that the industry as a whole needed to cover full production cost.¹⁵ Polysilicon fell so far that utilisation across the sector dropped to 40–50%, and Chinese polysilicon output fell 26.4% year on year while wafer output fell 12.3%.¹⁶

By 2025 the aggregate losses of China's listed solar companies were estimated to exceed RMB 50 billion.¹⁶ This was not a LONGi problem. It was a sector-wide destruction of capital.

What it did to LONGi's financials

The progression is stark and worth walking through slowly, because the shape of the deterioration tells you more than the levels.

2023 still looked fine on the surface — revenue of RMB 129.5 billion and net profit of RMB 10.75 billion — but the fourth quarter had already turned to loss. The cash statement was already flashing: operating cash flow of roughly RMB 8.1 billion against capital expenditure of RMB 9.3 billion meant free cash flow had gone negative while the company was still reporting a ten-billion-yuan profit. LONGi was, in other words, still spending into a boom that had ended.

2024 was the reckoning. Revenue fell 36.23% to RMB 82.58 billion and the company posted a net loss of RMB 8.618 billion — its first annual loss since 2013.¹⁷ The causes management gave were specific and, unusually, largely honest: a very low proportion of second-generation BC output, continued price and margin declines in PERC and TOPCon products, constrained capacity utilisation, larger asset impairments driven by technology iteration, and investment losses at polysilicon affiliates in which LONGi held equity stakes.¹⁷ The chairman described 2024 as the most difficult situation in the company's history.

Operating cash flow that year was negative RMB 4.72 billion, against capital expenditure of RMB 8.01 billion. That combination — cash out the door from operations, plus eight billion of capex — is what a company looks like when it is funding a technology transition in the teeth of a price war.

2025 was the stabilisation. Revenue fell a further 14.82% to RMB 70.347 billion, but the net loss narrowed by RMB 2.173 billion to RMB 6.42 billion, and — the number that actually mattered — operating cash flow turned positive at RMB 4.359 billion.¹⁴¹⁸ Total operating costs fell 8.72%; selling expenses fell 29.96% and administrative expenses fell 23.67%; inventory turnover improved by ten days; non-silicon wafer processing costs fell 29% year on year.¹⁴ R&D was held at RMB 4.3 billion.

Read that paragraph again, because it contains the single most important operational fact in the LONGi story right now: the company cut costs hard enough to restore positive operating cash flow while still losing money on the income statement, and it did so without cutting research. Whether that is discipline or denial depends entirely on whether the research pays off.

The human cost, and a credibility test

In November 2023 LONGi began cutting staff. In March 2024 Bloomberg reported, citing people familiar with the matter, that the company was planning to cut as much as 30% of a workforce that had peaked around 80,000.¹⁹²⁰ LONGi publicly denied the figure, stating that expected reductions were around 5% of total employees and calling the 30% figure false.²⁰

This is a small thing that is worth recording, because it is a data point on management communication under stress. The company's public number was 5%; the reported number was 30%; headcount disclosure since has not been granular enough to fully adjudicate the gap. Investors assessing management credibility should note that LONGi's instinct in a crisis was to issue a firm, low, public denial of a specific number rather than to explain the restructuring plan. That is a common corporate reflex. It is not a confidence-building one.

The peer scoreboard

Comparison is where this gets analytically useful, because it separates "the industry is terrible" from "this company is losing."

In the first half of 2026, LONGi guided to a net loss of RMB 3.4–3.8 billion, with recurring losses of RMB 3.7–4.2 billion. Tongwei — the polysilicon cost champion that expanded aggressively downstream into cells and modules and helped ignite the price war — guided to a loss of RMB 4.8–5.4 billion. TCL Zhonghuan, LONGi's principal rival in large-format wafers, guided to RMB 3.0–3.3 billion. JA Solar guided to RMB 2.4–2.9 billion, hit hard by the removal of export VAT rebates and rising trade barriers. Aiko Solar, the other BC believer, guided to RMB 680–790 million. And Trina Solar guided to a loss of only RMB 180–360 million, narrowing 88–94% year on year, with the second quarter expected near breakeven.²¹

That Trina number is the uncomfortable one. On these disclosures, the company nearest to breakeven in the first half of 2026 was not the technology leader. It was the company that focused on supply-chain standardisation. Meanwhile JinkoSolar, which had gone all-in on TOPCon, edged past LONGi to lead global module shipments in the first half of 2025 and finished 2025 effectively tied with LONGi in the 80–90 GW band, both well ahead of Trina and JA Solar at 60–70 GW.²²²³

So what does the crisis actually tell an investor? Three things. First, LONGi's cost reduction is real and measurable — the 29% fall in non-silicon wafer processing cost and the swing to positive operating cash flow are hard numbers, not narrative. Second, it is not yet enough: eleven consecutive quarterly losses is a long time to be wrong about a cycle. Third, and most importantly, LONGi's losses are not obviously better than the peer group's despite its technology differentiation. If BC were already delivering a premium, you would expect to see it in relative profitability by now. You do not.

Myth versus reality

Three consensus stories about this period deserve checking against the record, because all three are repeated confidently and none survives intact.

Myth: LONGi was the disciplined one that refused to over-expand into the boom. This was true in 2011 and is not true of the last cycle. In 2023 the company spent RMB 9.3 billion on capital expenditure against roughly RMB 8.1 billion of operating cash flow — it was outspending its own cash generation in the year the price collapse began — and spent a further RMB 8.0 billion in 2024 while operations consumed cash. LONGi expanded into the top of the cycle like everyone else. What distinguishes it from the weakest players is not that it avoided the mistake but that it entered the downturn with enough cash to survive making it.

Myth: the losses are purely an industry phenomenon and say nothing about LONGi specifically. Half right. The sector destroyed capital wholesale. But management's own explanation of the 2024 loss listed the low share of second-generation BC output and impairments from technology iteration alongside industry pricing — company-specific causes, disclosed by the company.¹⁷ Some of the pain was self-inflicted timing.

Myth: LONGi lost its crown to JinkoSolar. The shipment data does not support a decisive handover. Jinko took the lead in the first half of 2025 and the two finished the full year effectively level.²²²³ The more accurate statement is that LONGi's monopoly on the top spot ended while its scale did not. That distinction matters, because the bear case is not that LONGi is shrinking — it is that being the largest is no longer worth anything.

Which brings us to the bet itself.

↑ Back to Top

VI. The High-Stakes Technology Bet: BC vs. TOPCon & Management Assessment

In September 2023, with the industry still reporting record profits and every competitor pouring capital into TOPCon lines, LONGi's chairman stood up and told the market that TOPCon was a transitional technology and that back-contact cells would be the mainstream choice for the next five to ten years. It was the single most consequential statement in the company's recent history, and three years later the market still has not decided whether it was prescience or rationalisation.

First, the physics, in plain language

A conventional solar cell has thin metal fingers printed across its front face to collect the electrons that sunlight knocks loose. Those fingers are necessary and also self-defeating: metal is opaque, so every square millimetre of grid line is a square millimetre of silicon that never sees the sun. It is like installing skylights and then bolting a ladder across them.

TOPCon — tunnel oxide passivated contact — keeps the front grid but adds an ultra-thin oxide layer plus a doped polysilicon layer at the rear. That stack lets electrons tunnel through while blocking the recombination losses that waste energy at the metal contact. Its great commercial virtue is that a PERC line can be upgraded to TOPCon with modest additional capital. That is why the industry chose it: it was the cheap path.

Back contact takes the more radical route: move all the metal to the back. The front face becomes an uninterrupted sheet of silicon with zero shading. The cost is manufacturing complexity — you now need laser patterning to define alternating positive and negative regions on the rear, more process steps, tighter tolerances, and a redesigned interconnection scheme for stringing cells together.

The efficiency argument is real. Crystalline silicon has a hard theoretical ceiling near 29.4–29.6%. Back contact can approach it more closely than TOPCon can; industry technologists put BC's practical ceiling around 29.1% against roughly 28.7% for TOPCon, and commercial BC is expected to reach 27.5–28% cell efficiency over time.¹⁵²⁴ TOPCon's commercial progress, by contrast, has been slowed by the difficulty of passivating both faces well.²⁴

So the physics favours BC. The question was never the physics. The question was whether the extra cost of the process could be driven down fast enough to matter before the efficiency advantage was competed away — which is precisely the same question LONGi answered correctly for monocrystalline silicon twenty years ago, and which it has so far answered only partially for BC.

What actually happened

LONGi says it began working on BC in 2017. Its first-generation product, 复合钝化背接触 HPBC 1.0, went into mass production and encountered exactly the problems the sceptics predicted: yields below plan, a cost structure well above TOPCon, and limited traction in the utility-scale segment where price dominates. Management acknowledged the yield issues. But the commercial consequence was severe, and the 2024 results state it plainly — the very low proportion of second-generation BC output was one of the named causes of that year's RMB 8.6 billion loss, alongside impairments driven by technology iteration.¹⁷

Meanwhile the TOPCon fleet scaled. In 2025, global module production reached 669 GW, of which TOPCon was 516 GW — roughly 77% of the market.²⁵ LONGi had not merely chosen a different road; it had chosen a road that four-fifths of the world had declined to build.

The second generation is a genuinely different animal. HPBC 2.0, sold as Hi-MO 9, reached mainstream mass production at 650–660 W with peak power of 670 W and maximum module efficiency of 24.8%.¹⁴ By the end of 2025 LONGi had 46 GW of in-house HPBC 2.0 cell capacity plus 11 GW of collaborative capacity with partners.²⁶ In 2025 BC module sales reached 22.87 GW.¹⁴ At Intersolar Europe in 2026 the company launched the Hi-MO 9 Prime series at up to 680 W and 25.2% module efficiency, alongside a set of scenario-specific variants.²⁷

And the ramp in 2026 has been steep. In the first quarter, BC shipments hit 8.34 GW, or 66.1% of output, up from roughly a quarter a year earlier.²⁵ Across the first half, BC sales reached 19.55 GW — up 125% year on year — with the mix above 65% of module shipments.¹³

Testing the claim against the company's own record

Here is where the analysis has to get uncomfortable, because the bull case rests on a specific historical analogy and the analogy is weaker than it looks.

The claim: LONGi has an exceptional record of contrarian technology bets — mono over poly, diamond wire, downstream integration — and BC is the fourth in that sequence.

Test one: are these the same kind of bet? No. The mono bet was a cost-convergence bet: mono was already better, and LONGi wagered that engineering could remove its cost penalty. It won because the penalty went to zero and then negative. The BC bet is an efficiency-premium bet: BC costs more to make, and LONGi is wagering that customers will pay for the extra watts per square metre. Those are different mechanisms with different failure modes. A cost-convergence bet wins by definition once costs converge. A premium bet only wins if the buyer's willingness to pay persists — and in a market with 500-plus GW of TOPCon capacity desperate for volume, willingness to pay is exactly what erodes first.

Test two: what does the record say about LONGi's conversion of technical leads into money? Mixed, and getting less favourable. The diamond-wire achievement was converted into an enormous but temporary advantage that the company then diffused across the industry through its own supplier ecosystem. Continuous Czochralski research was launched in 2011 and discontinued by 2015 on competitive grounds, then restarted in limited form in 2018 — a reminder that not every LONGi research programme converted.⁴ HPBC 1.0 was a genuine commercial failure whose cost sits inside the 2024 impairments. The perovskite tandem programme has set world records repeatedly — 33.9% in November 2023, 34.85% certified by NREL in April 2025, 35.5% certified by ESTI in July 2026²⁸²⁹³⁰ — and has produced, so far, no commercial revenue whatsoever. Records are not products.

Test three: is the counter-positioning real? Partially, and less than claimed. 爱旭股份 Aiko Solar has been pursuing back-contact architecture independently and aggressively. More tellingly, in September 2025 LONGi and JinkoSolar settled a sprawling global patent war — cases across China, the United States, Europe, Japan and Australia, which Jinko had opened in January 2025 over TOPCon patents and LONGi had answered with a US counterclaim — with a cross-licensing arrangement covering selected core patents.³¹³²³³ Jinko had strengthened its position by acquiring 745 patents from LG Group in 2022, covering both TOPCon and back-contact technology.³¹ A cross-licence between the world's two largest module makers is not what a defensible cornered resource looks like. LONGi holds a substantial BC portfolio — 510 BC-related patents at end-2025, rising to about 630 by mid-2026 out of more than 4,100 granted patents overall¹⁴¹³ — but patents that are cross-licensed to your largest competitor bound the size of the moat.

Verdict on the claim: the history does not reject it, but it narrows it substantially. LONGi's genuine, repeatable, evidenced strength is manufacturing cost engineering — taking a known process and beating everyone else's cost curve. That strength is demonstrated across RCZ, diamond wire, and the 29% non-silicon cost reduction in 2025. What is not demonstrated is an ability to sustain a price premium for a differentiated product. The BC thesis requires the second capability, and the record supports mainly the first.

The promise on the table

Which is why one specific management statement matters more than anything else said in the last two years. At the April 2026 results briefing, Zhong Baoshen stated that BC product costs had reached broad parity with TOPCon in March, and that the company expected module profitability in the third quarter of 2026.²⁵

That is a falsifiable, dated, near-term commitment, and it is the right one to hold management to. It is being made against a backdrop where LONGi's capacity utilisation has run around 60%³⁴ and the company has just guided to a wider first-half loss than the prior year, attributing the shortfall to supply-demand imbalance, weak Chinese installations after a rush-installation base effect, insufficient grid absorption capacity, joint-venture investment losses and foreign-exchange losses on RMB appreciation.²¹

The people, and the governance questions

Li Zhenguo — founder, semiconductor physicist, and for a quarter-century the operational centre of the company — stepped down on 27 May 2025 as director, general manager and legal representative, moving to run the central research institute as chief technology officer. Zhong Baoshen took the general manager and legal representative roles alongside the chairmanship. Li's daughter, Li Shuxuan, was put forward as a non-independent director. The shares fell 2.7% on the day.[^3]

Read charitably, this is a founder returning to the bench he is best at, at the exact moment the company's survival depends on a technology transition. Read sceptically, it is a concentration of power in a single executive combined with the introduction of a family member to the board — a governance structure that a Western activist would question immediately, and one that reduces the number of independent checks precisely when strategy is contested.

Both readings are available. The evidence does not settle between them. What can be said factually is that Li retains effective control acting in concert with his wife Li Xiyan and co-founder Li Chunan; that he has pledged 211 million shares, about 2.78% of the company and 19.77% of his own holding, which is a modest but non-zero source of forced-selling risk if the stock keeps falling; and that Zhong Baoshen bought 6.13 million shares on the open market through 21 November 2025, roughly 0.08% of the company.³⁵³⁶

That last item is worth a moment. Insider buying during a drawdown is meaningful signal — but 0.08% of shares outstanding is a gesture, not a conviction trade. Investors should weigh it as such.

The shareholders who left

The other side of the register tells its own story. GL Ventures, the venture arm associated with Hillhouse, bought 226 million LONGi shares in December 2020 for roughly RMB 15.8 billion, around RMB 70 per share. At the February 2021 peak near RMB 125 the paper gain was around RMB 12.4 billion. By the first quarter of 2026, GL Ventures had exited the top-ten shareholder list, crystallising losses in the billions as the stock fell more than 80% from its high.³⁴ Separately, in November 2023 Hillhouse was placed under investigation by Chinese securities regulators for failing to disclose share sales as its LONGi holding crossed the 5% threshold, and was subsequently required to buy back shares.³⁷

At a May 2026 shareholder meeting, an individual investor told the board he had put RMB 15 million into LONGi and had RMB 9 million left, while comparable investments had doubled several times over, and asked directly whether the company's strategic positioning had been wrong. Zhong acknowledged that the share-price decline had caused investors losses and said he understood their feelings — while continuing to defend the BC strategy.²⁵

For investors, the honest summary is this: LONGi's management has been consistent, which is a virtue, and has been consistently wrong about timing, which is not. The narrative has not shifted opportunistically — the BC story told in 2023 is the BC story told in 2026, with the numbers improving each year — and that consistency deserves credit. But consistency is only a virtue if the destination is reached. The third quarter of 2026 module profitability claim is the first hard checkpoint.

↑ Back to Top

VII. Optionality & Hidden Business: Green Hydrogen (`隆基氢能 LONGi Hydrogen`)

Every large Chinese energy manufacturer eventually announces a hydrogen division, and most of them should be ignored. LONGi's deserves a paragraph or two more than that, but not many more — and the discipline of sizing it correctly is itself part of the analysis.

隆基氢能 LONGi Hydrogen — formally Xi'an LONGi Hydrogen Energy Technology — was established in 2021 to build large-scale 碱性电解槽 alkaline electrolysers.³

The strategic logic, which is genuinely sound

The problem LONGi is trying to solve is one it helped create. As solar penetration rises, midday generation increasingly exceeds midday demand. Grids curtail output; prices at the sunniest hours collapse toward zero. In China this shows up directly in LONGi's own results — the company has explicitly cited insufficient new-energy absorption capacity as a driver of weak installations.²¹

Electrolysis offers a physical answer: use the otherwise-worthless midday electricity to split water into hydrogen, then feed that hydrogen into industrial processes that electricity cannot reach — ammonia for fertiliser, methanol and synthetic fuels, direct reduction of iron for green steel, refinery hydrotreating. It converts a curtailment liability into a chemical feedstock, and it extends the addressable market for cheap solar electricity well beyond the power sector.

An alkaline electrolyser, mechanically, is not exotic: two electrodes in a caustic potassium hydroxide solution, separated by a diaphragm, with direct current forcing water molecules apart. It is century-old chemistry. The competitive variables are energy efficiency — how many kilowatt-hours per normal cubic metre of hydrogen — capital cost per megawatt, and reliability under the wildly variable input power that a solar farm actually delivers. That last one is where solar expertise plausibly transfers.

What it is actually worth today

LONGi's flagship products are 1,000 and 2,000 Nm³/h alkaline units, and the company has won meaningful domestic tenders — including 15 sets of 1,000 Nm³/h systems for the Da'an project in Jilin, representing 38.5% of that tender. It has begun exporting, shipping its first 5 MW alkaline system to Europe in January 2026 and supplying a green ammonia pilot in Uzbekistan. Manufacturing capacity was heading toward 3.5 GW as of October 2025, with company targets in the 5–10 GW range, and approval for a 1 GW alkaline factory.³⁸³⁹ Its principal domestic competitor is 阳光电源 Sungrow, which has been winning international orders in Oman, Kenya, Spain and Brazil from a comparable manufacturing base.

And the revenue contribution remains a low single-digit percentage of the group — a rounding error against a business that shipped 111 GW of wafers last year.

This is where the discipline matters. Five years after founding, hydrogen is still immaterial to LONGi's financial results. That is not a criticism of the technology; global green hydrogen demand has consistently disappointed against forecasts, and final investment decisions on large projects have repeatedly slipped. But it is a hard fact about conversion, and it should temper how much weight the optionality carries.

The relevant historical test is straightforward: how often has LONGi turned a technical milestone into revenue? The record is that it did so brilliantly once, with diamond wire and RCZ, over roughly five years, in its core process. Outside that core, the conversion record is thin — perovskite records with no product, HPBC 1.0 with poor yields, and now hydrogen at low single-digit revenue after five years. A reasonable investor should treat LONGi Hydrogen as a free option with a low probability-weighted value, not as a second leg of the business.

The same scepticism should be applied, immediately and without sentiment, to the newest expansion. In 2026 LONGi launched full-stack integrated solar-plus-storage solutions, targeting 6 GWh of energy storage system shipments for the year, and had signed cumulative storage orders above 3 GWh by mid-year.²⁶¹³ Zhong has described this as a strategic upgrade from photovoltaic leader to an integrated "solar-storage-hydrogen" energy solutions provider.⁴⁰

Storage is a large and fast-growing market. It is also a market with an entrenched incumbent structure, thin margins, and a cost curve driven by battery cell suppliers over whom LONGi has no control. A company losing money in its core business, midway through a capital-intensive technology transition, announcing entry into a new hardware category is precisely the pattern that a sceptical investor labels diworsification. It may well work. But it should be judged on delivered gross margin, not on strategy-slide adjacency — and it competes for the same management attention and capital as the BC ramp that the entire equity story depends on.

↑ Back to Top

VIII. 7 Powers Analysis & Business Lessons

Strip away the narrative and ask the structural question: what, if anything, protects LONGi's profits from competition when the cycle turns?

Scale Economies — real, but shared

LONGi's wafer operation is among the largest in the world, and gigawatt-scale slicing facilities genuinely spread fixed overhead thin. The 29% year-on-year reduction in non-silicon processing costs in 2025 is evidence that scale plus process discipline still yields something.¹⁴

The limitation is that TCL Zhonghuan operates at comparable scale in wafers, and JinkoSolar, Trina and JA Solar all operate at comparable scale in modules. Scale economics protect the top tier from the second tier. They do not protect the top tier from each other, which is exactly why four companies with world-class scale all lost money simultaneously in the first half of 2026.²¹

Process Power — the strongest of LONGi's powers, and still finite

This is where the company's genuine advantage lives. Two decades of accumulated knowledge in Czochralski thermal-field control, crucible management, ingot recharging, wafer thinning and now laser patterning for back-contact cells constitutes deep organisational know-how that cannot be bought as a machine. The evidence is the sustained cost gap LONGi has maintained through multiple cycles.

The counter-evidence is equally clear and comes from LONGi's own history: process advantages in solar diffuse through the shared equipment supply chain. The diamond-wire episode is the definitive proof. LONGi's edge was worth several extraordinary years and then became everyone's baseline. Investors should model process power in this industry as a lead-time advantage of one to three years, monetised through cost, not as a permanent structural moat.

Counter-Positioning — the claim, and the problem with it

The bull argument is elegant: incumbents with hundreds of gigawatts of freshly built TOPCon lines cannot pivot to back contact without writing off those assets, exactly as the polysilicon incumbents could not pivot to mono in 2015.

The historical parallel is seductive but structurally weaker than it appears, for three reasons.

First, the poly-to-mono transition destroyed incumbents because it was cost-driven — mono became cheaper, so there was no choice. BC is currently at best cost-parity with TOPCon on LONGi's own account, and only reached parity in March 2026.²⁵ A technology that is merely as cheap does not force a write-off; it permits waiting.

Second, the incumbents are already inside the tent. The Jinko cross-licence covers back-contact-relevant intellectual property, and Jinko holds LG-derived BC patents.³¹³³ Aiko is a committed BC producer. Counter-positioning requires that competitors cannot or will not respond. Several already have.

Third, capacity arithmetic. LONGi has built roughly 60 GW of BC capacity against a global market that produced 669 GW of modules in 2025.²⁵ Even a full LONGi conversion leaves BC a minority technology for years. Ping An Securities projected BC share above 10% in 2026; Zhong himself has projected 50% by 2027 or 2028.²⁵²⁴ The gap between those two forecasts is the entire investment debate.

Cornered Resource — weakest link

The BC patent portfolio — 630 BC-related grants within more than 4,100 total by mid-2026¹³ — is substantial. It is also, in part, cross-licensed to LONGi's largest competitor. A cornered resource that you have contractually shared with the other market leader is not cornered. It is a bargaining chip that has already been played.

Brand and Switching Costs — thin, by the nature of the product

Solar modules are bought by developers running financial models. Bankability matters — a Tier 1 rating and a balance sheet capable of honouring a twenty-five-year warranty are real assets, and LONGi's are among the strongest in the sector. But there are no switching costs worth the name. A developer can specify a different module for the next project with no penalty beyond re-running a layout.

The two lessons

The commodity technology paradox. Global solar installations have grown from a rounding error to hundreds of gigawatts a year. Almost none of that value accrued to module manufacturers. When a product is undifferentiated at the point of use, when capital intensity is low enough for regional governments to subsidise entry, and when the equipment that embodies the process innovation is sold by third parties to all comers, then demand growth funds capacity growth rather than margin. Investors who reasoned from "solar demand will explode" to "solar manufacturers will compound" were right about the premise and wrong about the conclusion for a decade running.

Value migration is real and it moves fast. In 2010 the profit pool sat in polysilicon refining. By 2016 it had moved to wafering, which is what made LONGi. By 2020 it had moved partly to cell technology and brand distribution. Today, arguably, it sits with equipment vendors and with whoever controls grid interconnection. Each migration destroyed the fortunes of companies that had optimised for the previous location. LONGi's BC bet is, in essence, an attempt to move the profit pool back to cell technology by force. Whether that is possible is the question the next two years will answer.

↑ Back to Top

IX. Bear vs. Bull Case, Risk Radar, & KPIs

Set the two arguments against each other properly, and the debate resolves into a small number of testable propositions.

The industry backdrop, through Porter

Rivalry is extreme and structural. Dozens of scaled producers, near-identical products, high fixed costs, and — critically — local-government incentives that keep marginal capacity alive past the point where economics would kill it. The 反内卷 anti-involution campaign is Beijing's explicit attempt to fix this. It has had partial success: eight leading polysilicon producers publicly pledged not to sell below cost benchmarks⁴¹, output fell sharply, and prices recovered, with n-type recharge polysilicon back near RMB 59,200 per tonne in early 2026 and module quotes climbing to RMB 0.70 per watt or above.⁴²¹⁵ But the campaign's most ambitious move — a producer consortium to buy and retire roughly a third of polysilicon capacity — was suspended by China's antitrust regulator in January 2026 on monopoly grounds.¹⁶ Administrative discipline in an industry with fast technology iteration and powerful scale economics is a patch, not a cure.

Supplier power is moderate and rising in an unexpected place: silver. Metallisation paste is one of the few inputs the industry cannot substitute freely, the solar sector already consumes more than a quarter of annual silver production, and rising silver prices were explicitly cited as a cost driver behind LONGi's first-quarter 2026 losses on some utility-scale orders.²⁵ Both BC and TOPCon roadmaps depend on reducing or eliminating silver.

Buyer power is very high. Utility-scale developers run auctions, hold no loyalty, and have hundreds of gigawatts of eager supply to choose from.

Substitutes and new entrants — the entry barrier for module assembly is capital, not capability, which is the root of the whole problem.

The bear case

The price war outlasts the balance sheet. Eleven consecutive quarterly losses have already consumed roughly a quarter of book equity — total shareholders' equity fell from RMB 70.5 billion at end-2023 to RMB 53.8 billion at end-2025. The asset-liability ratio stood at 64.43%.²⁶ Cash remains substantial at more than RMB 48.6 billion in monetary funds at mid-2026¹³, and LONGi's balance sheet is genuinely among the healthiest in the sector — but "healthiest in a sector that lost RMB 50 billion in a year" is a relative statement. Two more years at the current burn changes the conversation.

It is worth stress-testing the "balance sheet resilience" claim directly rather than accepting it. LONGi has raised external capital repeatedly through its history — a RMB 7 billion convertible bond issue to fund cell capacity, and the December 2020 placement that brought GL Ventures in at RMB 70 per share.³⁴ Free cash flow has been negative in each of 2023, 2024 and 2025 on the reported operating cash flow and capital expenditure. The company is not distressed. But the record does not support a characterisation of a business that has never needed the capital markets, and with the shares near RMB 12 against that RMB 70 placement price, equity is an expensive funding source today.

BC does not earn its premium. This is the core bear argument. If BC reaches cost parity but not cost advantage, and the efficiency premium is competed away by a TOPCon fleet running at low utilisation and pricing to marginal cost, then LONGi will have spent several years and billions of yuan of capital to arrive at the same commodity economics as everyone else — with a smaller installed base and a technology the rest of the industry can now license.

Trade and policy walls keep rising. The April 2025 US determinations closed the Southeast Asian route¹¹; UFLPA enforcement continues to constrain the highest-margin market¹⁰; the European Union has shown it will use the Foreign Subsidies Regulation against Chinese solar bidders, opening an in-depth investigation on 3 April 2024 into a consortium including LONGi's German subsidiary and Romanian firm Enevo Group over a 455 MW Romanian solar park, after which LONGi withdrew from the tender and the case was closed.⁴³⁴⁴ Withdrawal ended the investigation without a finding; it also demonstrated that a European public-procurement channel can be closed to LONGi at administrative speed. The direction of travel in Brussels compounds this: the EU's Net Zero Industry Act framework attaches non-price resilience and sustainability criteria to public renewable auctions, which is a polite way of describing local-content preference. None of these measures bans Chinese modules. What they do is convert LONGi's two highest-margin export markets from open price competition — the game it wins — into administered procurement, where the winning variable is where a factory sits rather than what a module costs per kilowatt-hour. That is a structural repricing of the export franchise, and it is largely outside management's control.

Execution and attention risk. A simultaneous BC ramp, a storage entry, a hydrogen business and a perovskite research programme, run by a management team that has just concentrated the chairman, general manager and legal representative roles in one person, is a lot of surface area.

The bull case

The technology gap is finally showing up in the mix. BC at 66.1% of first-quarter output and 19.55 GW in the first half — up 125% — is not a pilot. It is a completed industrial transition of the majority of a 80 GW module business in under two years, and the pace of it is genuine evidence of manufacturing capability.²⁵¹³

Cost discipline is working where it can be measured. Non-silicon wafer processing cost down 29%, selling expenses down 30%, administrative expenses down 24%, inventory turnover ten days better, and operating cash flow restored to positive RMB 4.359 billion while still loss-making.¹⁴¹⁸ That is a management team executing a cost programme competently.

Consolidation may finally be arriving with teeth. Three mandatory national standards take effect on 1 January 2027 and are expected to eliminate roughly 30% of outdated capacity; on one assessment, about 47% of existing capacity among the top eleven manufacturers fails the Tier-3 energy efficiency requirement.²⁵ Regulatory-forced retirement succeeds where price signals have failed, because it is not optional. LONGi, with newer capacity and a positive cash position, is better placed than most to survive that cull and to buy distressed assets if it chooses.

The demand backdrop remains intact. Chinese module exports reached 267.6 GW in 2025, up 13%, and LONGi's overseas module shipments rose more than 26% in the first half of 2026, with the Americas up over 36% and Europe up over 34%.²³¹³ The end market is not the problem. It never was.

Weighing it

The bull and bear cases do not deserve equal weight, and it is worth saying which way the evidence currently leans.

On the operational claims, the evidence supports LONGi. Cost reduction, cash-flow restoration and the speed of the BC mix shift are documented, dated and substantial. Management said it would convert the fleet to BC and it converted the fleet to BC.

On the economic claim — that BC will generate a durable premium — the evidence is not yet there, and the company's own history argues for caution rather than confidence. LONGi's demonstrated superpower is cost engineering, not premium capture; the one time it created a decisive process advantage, it diffused that advantage across the industry within a few years; and its largest competitor now holds a cross-licence to relevant intellectual property. The correct posture is that the BC thesis is unproven rather than refuted — narrowed from "BC will restore LONGi's structural advantage" to the smaller and more testable "BC can restore LONGi to profitability ahead of the TOPCon field."

The activist question follows naturally. A sceptical long/short investor would ask why a company that has lost money for eleven straight quarters is opening a new hardware business line in energy storage; why the founder's departure from executive roles coincided with a family board appointment; why the response to reports of a 30% workforce reduction was a public denial rather than a restructuring plan; and whether the RMB 6.255 billion of BC projects still in progress will earn their cost of capital in a market where BC may hold only low-double-digit share by the end of 2026.³⁴²⁵ Those are fair questions, and none of them has been fully answered.

The KPIs that will settle it

Three numbers matter more than everything else. Investors should track them; there is no need to model anything more elaborate.

1. BC module gross margin — specifically, whether HPBC 2.0 modules turn a positive product-level margin, and when. Management committed publicly to module profitability in the third quarter of 2026 following cost parity with TOPCon in March. This is the single cleanest test of the entire thesis, it is dated, and it is falsifiable. Watch shipment mix and blended module gross margin together: mix above 65% with margin still negative would mean parity was achieved on cost but not on price.

2. Non-silicon wafer processing cost per watt, tracked against TCL Zhonghuan. This is the direct measure of LONGi's one genuinely demonstrated power. If the gap holds or widens, process power is intact and the company will outlast weaker rivals regardless of how the BC debate resolves. If it closes, the most durable part of the business is eroding.

3. Operating cash flow and the free cash flow gap. Operating cash flow turned positive in 2025 while the income statement stayed deeply negative. The question for the next several quarters is whether that holds and whether capital expenditure on the BC build-out falls enough to close the free cash flow deficit. Cash generation, not reported profit, determines how many more quarters of price war LONGi can absorb — and therefore whether it is a consolidator or a candidate for consolidation when the cycle finally turns.

Everything else in this story — the Lanzhou statue, the diamond wire, the world records, the strategy slides — is prologue to those three numbers.

LONGi Green Energy Technology Co., Ltd.

Table of Contents

LONGi Green Energy: The Photovoltaic Titan and the Solar Supercycle

I. Introduction & Episode Roadmap

II. The Lanzhou Pact & Monocrystalline Conviction (2000–2011)

The consensus they chose to ignore

The LCOE argument

Surviving the first great bust

III. The Twin Technological Breakthroughs: RCZ & Diamond Wire (2012–2017)

Act one: teaching the crucible to work overtime

Act two: cutting silicon with diamonds

The tipping point, and the moat LONGi gave away

IV. Downstream Integration: The LERRI Acquisition & Module Dominance (2014–2020)

Buying the smallest possible key

Building an offshore hedge

V. Financial Structure, Segment Breakdown, & The Overcapacity Crisis (2021–2025)

How the segments actually work

The anatomy of an overcapacity catastrophe

What it did to LONGi's financials

The human cost, and a credibility test

The peer scoreboard

Myth versus reality

VI. The High-Stakes Technology Bet: BC vs. TOPCon & Management Assessment

First, the physics, in plain language

What actually happened

Testing the claim against the company's own record

The promise on the table

The people, and the governance questions

The shareholders who left

VII. Optionality & Hidden Business: Green Hydrogen (隆基氢能 LONGi Hydrogen)

The strategic logic, which is genuinely sound

What it is actually worth today

VIII. 7 Powers Analysis & Business Lessons

Scale Economies — real, but shared

Process Power — the strongest of LONGi's powers, and still finite

Counter-Positioning — the claim, and the problem with it

Cornered Resource — weakest link

Brand and Switching Costs — thin, by the nature of the product

The two lessons

IX. Bear vs. Bull Case, Risk Radar, & KPIs

The industry backdrop, through Porter

The bear case

The bull case

Weighing it

The KPIs that will settle it

References

VII. Optionality & Hidden Business: Green Hydrogen (`隆基氢能 LONGi Hydrogen`)