Choosing the slower answer, four times

Choosing a provincial state investor over a higher private valuation

In early 2022 Qianyue needed about RMB 2.5 billion of equity for its next seven stations. Our bank loans cost about 5.6 per cent, several backed personally by our founder. A consortium of growth funds offered RMB 2.6 billion for 26 per cent, valuing us at RMB 10 billion, with a right to make us buy back their shares at an 8 per cent annual return if we had not listed by the end of 2025. Mingquan Energy Investment Group, a provincial state-owned investor, offered the same sum for 34 per cent, three board seats and no buy-back right.

The RMB 2.35 billion valuation gap fell on our founder, our existing investors and the 180 employees in our shareholding platform. Fang Dezhi, our founder and chairman, preferred the private offer, as did the existing investors. A state shareholder, they argued, would slow decisions and make us look like a policy instrument to developers competing with Mingquan's own generators.

I recommended Mingquan. The buy-back clause decided it: our revenue rested on provincial rules that could change, and I did not want an obligation above RMB 3.3 billion falling due in those same years. I also expected banks to lend more cheaply against stations with a patient, solvent owner behind them. I gave up valuation, three seats and some speed.

Negotiation took four months. Mingquan gained approval rights over budgets, borrowing above a ceiling, disposals and related-party transactions; tenders, hiring, station design and pricing stayed with management. Any contract with a Mingquan company would be decided by the directors Mingquan had not nominated, on an independent price benchmark. The founder agreed once he gained a second board nominee.

Mingquan invested in September 2022. By 2024 new project loans cost 3.7 per cent and lenders no longer sought the founder's personal backing. We added 3.0 GWh of stations from 2023 to 2025, and when we withdrew our listing application in 2024 no investor could force a sale. Approvals are slower: in 2023 a Mingquan director's objection held up our purchase of a part-built station for three months, and a rival bought it.

I would have written the conflict rules for Mingquan's generators into the investment agreement itself rather than a later board policy, and I would have explained the valuation trade-off to our employee shareholders in person before the vote.

Holding a price floor through the 2024 price war

By the spring of 2024 a standard two-hour storage system sold in China for around RMB 0.6 per watt-hour, half its price at the start of 2023. Cell makers with idle lines sold below full cost. Our Zibo plant had just added a third line, lifting capacity to 12 GWh a year.

In May 2024 a central state-owned generator opened a 1.6 GWh framework tender, our largest ever. Our head of domestic sales said we had to win it at almost any price: it would fill the new line for six months, keep 310 contract production workers employed and hold our place on the generator's approved list. Our founder agreed. Our cost for a two-hour system, including a reserve for fifteen years of warranty and service drawn from our own fleet's data, was about RMB 0.55 per watt-hour.

I set a floor of RMB 0.58 and told the board we would not bid below it, at home or abroad, until cell prices or our costs moved. The tender cleared near RMB 0.47. We lost it, and several smaller tenders that year. I gave up volume, a full plant and, as it turned out, our head of domestic sales, who joined a competitor in October.

We let the 310 contract agreements lapse, froze hiring and cut senior pay by a fifth for a year, mine first. Sales effort moved to exports and our own stations, and every declined tender went to the board with our reasons.

Domestic system revenue fell from RMB 3.4 billion in 2023 to RMB 1.9 billion in 2024; group revenue fell from RMB 5.9 billion to RMB 4.8 billion, and the plant ran at 46 per cent. Integration gross margin held near 10 per cent, warranty claims stayed inside our reserve and exports grew. Revenue recovered to RMB 5.45 billion in 2025, and two buyers that had chosen cheaper suppliers came back after trouble with their stations.

The floor is now board policy, reviewed quarterly, and sales incentives pay on contribution after the warranty reserve, not on gigawatt-hours shipped.

I would not have backed the third line; its forecasts assumed a market every competitor was building for. And I would have explained the floor to the contract workers' supervisors myself before the cuts.

Rebuilding station revenue after paired storage stopped being a condition of connection

Early in 2025 a central pricing reform moved new wind and solar output towards market prices and ended storage as a precondition for approval or grid connection. In 2024, 56 per cent of our RMB 1.18 billion of station revenue had come from leasing capacity to developers who needed it for provincial pairing requirements, and leases carrying about 40 per cent of that income would expire by the end of 2026.

Without leases, our stations had to earn from the Shandong spot market, capacity compensation and ancillary services: income we had treated as secondary and staffed with six people. Three stations totalling 900 MWh were being built on the old assumptions.

Three proposals reached me. Sales wanted to halve lease prices to keep developers, which would have held revenue for a year or two. Our chief financial officer wanted to sell the two Inner Mongolia stations at once. Mingquan's directors wanted the new stations finished, partly because the province's storage plan counted them. I chose a fourth course: renew leases only near the market value of the capacity, build a trading capability, pause construction and test every station against what it could earn. That meant admitting some stations were worth less than we had said.

We recruited a trading and forecasting desk of 24, built our own forecasts of solar output, load and congestion, and bid every Shandong station daily. Three large developers moved from pure leases to multi-year agreements with a lower fixed fee plus a share of trading margin. We paused the three stations after spending RMB 410 million and redesigned two for four-hour duration.

Lease income fell by about a quarter in 2025, yet station revenue rose to RMB 1.32 billion, and trading and compensation reached 71 per cent of it in the first half of 2026. The Inner Mongolia stations, sited for cheap connection rather than grid need, earned far less than their leases had paid; we impaired them by RMB 214 million. Two developers cancelled rather than renegotiate.

Each station now carries a quarterly statement against its investment case, and new stations are approved on trading forecasts, not lease commitments.

I would have built the trading desk in 2022, when the spot market first rewarded it, rather than treating leases as annuities, and told Mingquan's directors sooner that a provincial plan counting our capacity was no reason to build it.

Halting two stations after a fire during commissioning

On 17 August 2023, during commissioning of a 200 MWh station near Dongying, a battery module in one container overheated, vented gas and ignited. The suppression system operated and the local fire service attended. The container was destroyed, the fire did not spread and no one was injured. We reported the event to the local emergency management and energy authorities that day.

Our investigation, with an external laboratory, traced it to defective cells from one supplier's production batch and to battery management thresholds set more loosely than those cells needed. The same batch had gone into a second 200 MWh station near Binzhou, due to energise the next month. Both stations were leased in advance to developers who needed them for pairing obligations before their own connection dates.

Our project director proposed replacing the burnt modules, tightening thresholds in software and energising both stations on schedule. The supplier said the batch's defect rate did not justify more, and two developer customers pressed hard to keep their dates. Pei Hanlin, our head of safety and quality, wanted every module from the batch tested before either station was energised.

I backed him and halted both stations, accepting missed dates, penalties and the full cost before we knew whether the supplier would share it. We tested every module from the batch, replaced 31 per cent of the cells in the two stations and retrofitted module-level gas detection, explosion venting and wider container spacing.

The work cost RMB 168 million; the supplier reimbursed RMB 61 million after seven months of negotiation, and we paid RMB 22 million in late-capacity penalties. Both stations energised five months late, in January 2024. We have had no thermal event since across 5.6 GWh of operating storage, and teardown sampling of every incoming cell batch, introduced that autumn, has rejected four batches from three suppliers.

Our head of safety can now stop any energisation without my approval, and a thermal event anywhere triggers inspection of every station built from the same cell batch.

I would have begun teardown sampling when cell prices started falling in early 2023, before a fire showed us why. And I would have published our findings sooner: customers and lenders had them within a fortnight, other operators only a year later.