Where is Kaishan's profit growth coming from?

Between January 16 and 19, 2026, the controlling shareholder of Kaishan Group (300257.SZ) unloaded 9.94 million shares at ¥18 apiece. That's ¥179 million in cash. In the filing with the Shenzhen Stock Exchange, the reason given was "to support the Kenya green ammonia fertilizer project."

Routine stuff for a Chinese mid-cap. Except that on November 7, 2024, about ten weeks earlier, Kaishan had filed a separate announcement disclosing that the final $57 million EPC payment from the Kenya Sosian Menengai geothermal project had been received in full, along with $3.16 million in interest. The Kenya receivable was zero. Everything collected. (Source: Jiemian, Nov 7 2024)

So the question is basic: why sell shares to fund a project in a country where you just finished collecting every dollar owed to you?

We don't know the answer. Maybe there's a new ammonia project in Kenya that requires fresh capital. Maybe the language in the filing was imprecise. Maybe it's just easier to write "Kenya" than to explain the real reason. But this is the kind of gap between what a company says and what the filings show that sends you down a rabbit hole. We went down it.

What we found at the bottom was not a balance sheet problem. It was worse.

Five Dead in the Rice Paddies

Kaishan's showpiece is SMGP, or PT Sorik Marapi Geothermal Power, a 240 MW geothermal complex on the flanks of an active volcano in North Sumatra, Indonesia. Kaishan's subsidiary KS Orka acquired a 95% stake in 2016. By 2023, the plant was generating nearly a million megawatt-hours of electricity and $83.6 million in annual revenue, sold to Indonesia's state utility under a 32-year take-or-pay contract at roughly 8.1 cents per kWh. In the annual report, it reads like a textbook clean energy success.

On January 25, 2021, workers at SMGP opened well T-02 to commission Unit 2 of the power plant. Hydrogen sulfide gas, the rotten-eggs compound that kills in concentrated doses, blasted out of the well and drifted over the surrounding farmland.

A woman was in the rice fields, 150 meters from the well, with her two children. The children were three and five years old. All three died. Two other villagers also died. Nearly 40 people were hospitalized. (UCA News, Feb 15 2021)

The only sanction: a temporary suspension of operations. Within three months, everything was back online.

In Kaishan's 2021 semi-annual report, this event appears as a ¥30 million profit impact and a paragraph about the construction timeline being delayed. It mentions five deaths with the kind of language you'd use to describe a supply chain disruption.

It Kept Happening

Here's what makes the SMGP story so difficult to dismiss as a one-off industrial accident.

In March 2022, another H₂S leak sent 58 people to the hospital. A month later, a 30-meter mud eruption near the site hospitalized 21 more. In September 2022, it happened twice in one month, both from the same well pad T-11, and Jatam, the Indonesian Mining Advocacy Network, counted 87 victims from the second incident alone. SMGP's corporate communications department said no H₂S was detected by their fixed gas monitors. The 87 people vomiting and fainting in two neighboring villages apparently had some other explanation.

In February 2023, well testing poisoned over 120 residents. In February 2024, it happened again. Over 100 people across two villages fell ill after SMGP opened a borehole. BBC News Indonesia covered it.

Then in April 2025, something different. Twenty-one hot mud eruption points appeared across farmland in Roburan Dolok village, some as close as 10 to 15 meters from SMGP drilling sites and 317 meters from houses. Over five hectares of rubber, candlenut, cocoa, and rice paddies were destroyed or rendered infertile. Walhi (Friends of the Earth Indonesia) documented it. BNPB, the national disaster agency, confirmed 15 eruption points across 5 locations. SMGP's position: this is a "natural phenomenon." (Indonesia Business Post, April 2025)

We went through every Kaishan annual report and semi-annual report from 2021 to 2025. The January 2021 incident gets a brief mention in the risk section. The subsequent poisonings, the mud eruptions, the ongoing civil society campaigns, the Indonesian government's "maloperation" finding: none of it appears in the investor-facing documents in any meaningful way. It's not hidden; it's just not there. If you only read the filings, SMGP is a triumph of Chinese engineering.

Who Bought the Green Bonds

This is where the story gets genuinely strange.

In 2024, Kaishan issued green bonds for the Indonesian geothermal business. The semi-annual report celebrates the transaction: "international top-tier investment institutions became bondholders, representing international market recognition of the company's projects and leadership team." The bonds were meant to partially finance expansion of the same SMGP complex.

We couldn't identify the specific buyers from the public filings. But "international top-tier investment institutions" buying "green bonds" for a "geothermal project" suggests ESG-screened capital, the kind of money that supposedly goes through rigorous environmental and social due diligence before deploying.

We're not sure what to make of this, honestly. ESG frameworks are good at screening for carbon intensity and bad at screening for hydrogen sulfide in rice paddies. The SMGP carbon numbers are genuinely excellent. Geothermal is baseload clean energy, and every MWh from Sorik Marapi replaces coal. But the S in ESG stands for something, and this project has a body count.

The Real Balance Sheet

Step back from Indonesia for a moment and look at the company as a whole.

Kaishan's 2024 numbers: ¥42.35 billion revenue, up a paltry 1.6%. Net profit ¥3.2 billion, down 26%, missing consensus by about ¥340 million. The profit drop came from three things piling up at once. The Kenya EPC project was done, so that construction revenue vanished. Selling and administrative expenses grew. Credit impairment provisions went up.

The balance sheet is tighter than the income statement suggests. Current ratio: 0.92, which means current liabilities exceed current assets. Interest-bearing debt is 48% of total assets. Accounts receivable is 2.8 times net profit. Cash covers 57% of current liabilities. For a company promising investors that profits will more than triple from ¥320 million to ¥950 million between 2024 and 2027, this is a balance sheet with no margin for error.

And then there's the margin question that nobody seems to want to discuss plainly. Kaishan's operating margin runs about 8%. Atlas Copco's is around 21%, Ingersoll Rand's around 25%. That's not a gap. It's a chasm. For every dollar of compressor revenue, Kaishan earns roughly a third of the profit that its Western competitors do. Part of that is structural: price competition in China is vicious. Part of it is that Kaishan's service revenue, the high-margin aftermarket business that makes Atlas Copco so profitable, barely cracks 10% of sales versus 35 to 40% for the Western majors.

But the bigger issue is where the compressor profits go. Atlas Copco did 29 acquisitions in 2024 and 2025. Ingersoll Rand spent €160 million on Termomeccanica in Italy. They pour compressor profits back into buying more compressor businesses, which generate more profit, which fund more acquisitions. Kaishan takes the cash from selling screw compressors in China and Alabama and funnels it into geothermal wells in Sumatra and Nevada. Total assets grew 20% in a year, to ¥170.78 billion, driven mostly by geothermal investment. The compressor division is basically the ATM for a geothermal startup.

That last row is the one we keep coming back to. It's a legitimate strategic bet. Geothermal PPAs throw off cash for 32 years, and if American data center demand pushes PPA prices toward $100/MWh, the math changes dramatically. But every year the compressor business doesn't grow fast enough, and every year a geothermal well doesn't produce on schedule or produces H₂S instead of steam, the financial architecture gets more fragile. The current ratio is already below 1.

The Controlling Shareholder's Margin Account

There's a line in the 2025 semi-annual report that you'd miss if you weren't looking for it. Of the controlling shareholder's 566 million shares, 70 million sit in a "credit trading guarantee securities account" (客户信用交易担保证券账户) at CITIC Securities. That's a margin collateral account. About 12% of the controlling stake is being used as collateral for leveraged positions.

Compare that with Atlas Copco's register, which is a who's-who of global pension and sovereign wealth funds. Capital market credibility is itself a competitive asset, and it's one that Kaishan can't buy with an acquisition or build with a new product line.

Buffett's Leftovers

One more thing. In February 2021, the same month that five people died at SMGP, Kaishan announced it had acquired the Fish Lake geothermal project in Nevada from Berkshire Hathaway Energy for $5.1 million. The press release noted that the site's assets were "valued at $15 million, much higher than the contract price."

This was widely reported as a sweetheart deal. Look deeper and the picture shifts. BHE operates 10 geothermal plants at the Salton Sea in California with 345 MW of combined capacity. Their real play at those plants is lithium extraction from geothermal brine, a bet worth potentially billions. Fish Lake was a tiny undeveloped project in rural Nevada with five wells and no power plant. BHE was cleaning out the junk drawer. Kaishan was buying what the richest energy company in America had decided wasn't worth developing.

The fourth U.S. acquisition, Bottle Rock in Lake County, California, followed a similar pattern. Twenty-three existing wells, but only an estimated 7.5 MW of sustainable generation using Kaishan's ORC equipment. The previous operator was selling. These are recycled assets from the bottom of the geothermal industry's depth chart.

What's Actually Good

After 3,000 words of this, we should talk about the compressor business, which is genuinely impressive in ways that have nothing to do with geothermal holes in the ground.

KCA, the Alabama subsidiary, holds about 9% of the North American screw compressor market. For a Chinese brand competing against Atlas Copco, Ingersoll Rand, Sullair, Quincy, companies with decades of installed base and service networks, that's a serious number. The "Y-series" screw airend was developed at a Seattle R&D center and has passed China's national first-tier energy efficiency certification across the full range. That's not cheap-box selling. That's real engineering.

LMF in Austria is a different kind of story. Acquired in 2016 when it was doing €36 million, it hit €76.3 million in 2024 and management expects 10% revenue growth and 40% profit growth in 2025. The growth driver is weirdly specific: European gas utilities in Belgium, France, Germany, Denmark, Spain are ordering mobile high-pressure compressors for pipeline purging work at rates LMF has never seen. Also dominates a global niche in offshore exploration compressors. One hundred seventy years of piston compressor heritage. Not something you can replicate.

In specialty niches, some startling numbers. Kaishan holds over 60% of the Chinese market for argon recovery compressors in the photovoltaic industry, with 100+ units delivered. First company to supply both crude-argon and high-purity-argon isothermal centrifugal compressors with sub-0.1% leakage rates. The hydrogen compression unit is operational. Magnetic levitation blower and vacuum product lines are scaling.

There's even an interesting long-shot in marine air lubrication, hull air-bubble systems that reduce ship drag by 8% or more. The compressed air requirements happen to match screw compressor output profiles. Addressable market: roughly 30,000 large commercial vessels worldwide. Speculative, but the kind of thing that could matter in five years.

All of this is real. It's the reason the stock has any following at all, and the reason brokers write their ¥400M → ¥600M → ¥950M profit trajectory reports.

The Gap

We've read the broker reports. We've read the investor day transcript. We've read the annual reports in both Chinese and English. Everyone tells the same version: visionary founder, dual-engine model, geothermal optionality, undervalued relative to Ormat Technologies.

It's possible that the SMGP incidents are genuinely isolated operational errors in a dangerous industry. Geothermal development involves drilling into volcanic systems, and H₂S risk is inherent and well-known. It's possible that SMGP has improved its protocols since each incident. It's possible that Jatam's advocacy represents an adversarial position rather than an objective assessment.

It's also true that the incidents didn't stop. They kept happening: 2021, 2022, 2022 again, 2023, 2024, and the mud eruptions in 2025. The Indonesian government's own investigation found maloperation. A civil society coalition organized a public protest at the country's biggest geothermal conference specifically targeting this project. And not one word of any of this appears in how Kaishan presents itself to capital markets.

The compressor engineers in Quzhou, the LMF machinists in Austria, the KCA salespeople in Alabama: their work is good work. Whether it's good enough work to carry a company whose profit growth depends on an Indonesian geothermal complex ringed by traumatized villages is a question that each reader has to sit with.

We just wanted to make sure you had the information.

We sent questions to Kaishan's investor relations department on February 7. No response as of publication.