Where judgement is made in a storage company

I judge each of our 21 stations on what it earns per kilowatt of capacity per year, net of degradation and the power it consumes itself, and on how reliably it is available when the system operator calls. Before 2025 most of our station revenue came from leasing capacity to wind and solar developers who needed it to meet provincial pairing requirements. Those leases paid for presence, not performance. When the 2025 reform removed storage as a condition of grid connection, I chose to rebuild revenue around the Shandong spot market and capacity compensation rather than cut lease prices to hold customers. That required a trading desk, better forecasting of solar output and congestion, and a revised capital plan: we paused three stations totalling 900 MWh, impaired two weaker ones and began designing four-hour stations where evening spreads justify them. Each station now has a quarterly statement comparing its earnings with its original investment case.

Most battery fires begin long before a station is energised: in a cell with an internal defect, a module assembled with poorly torqued busbars, or a battery management system whose thresholds were set for a different cell. Since our commissioning fire in August 2023, our safety programme starts at incoming inspection. Every cell batch is traced to its production line and date, and a sample from each batch is torn down and examined in our own laboratory before acceptance. We added module-level gas detection, explosion venting and wider container spacing to our standard design, and every station's thermal data is monitored centrally around the clock. A thermal event anywhere triggers inspection of every station built with the same cell batch. Safety decisions rest with our head of safety and quality, who can stop an energisation without my approval. My part is to fund the programme and defend its cost to the board.

Equipment prices in Chinese storage fell by more than half between early 2023 and late 2024 as cell and integration capacity outran demand. In that market the most consequential decision a chief executive makes is which business to refuse. We set a price floor for system sales built from three parts: full production cost, a reserve for fifteen years of warranty and service obligations calculated from our own fleet's degradation data, and a modest margin. We do not bid below it, at home or abroad. The floor cost us domestic volume and left our Zibo plant at 46 per cent utilisation in 2024. It also held integration gross margin at about 10 per cent and kept us able to honour our warranties. I review the floor with our chief financial officer every quarter against cell prices and claims experience, and every tender we decline goes to the board with the reason.

Qianyue was privately founded but has a provincial state-owned investor as its largest shareholder, and I have learnt to run a company that answers to both cultures. When we raised RMB 2.6 billion in 2022, I negotiated a list of reserved matters that keeps tenders, hiring, design and pricing with management while giving Mingquan Energy Investment Group the reporting and approvals a state investor needs. Its presence lowered our project-debt cost and lengthened our loan tenors. It also brought procedures, such as formal benchmarking of every contract with its group companies, that slow us down and that I defend because they protect our other shareholders. Our project debt stood at RMB 6.1 billion at the end of 2025, almost all of it from domestic banks lending against individual stations. In 2024 I recommended withdrawing our listing application, because I could not defend a valuation while our revenue model was changing beneath us.

Exports brought in RMB 1.57 billion of our 2025 revenue, from customers in Australia, the Gulf and Chile. Abroad, the buyers who matter are usually financed by banks whose engineers read our warranty and service plan before they look at our price. We therefore price the warranty explicitly and hold a reserve against it, and we do not sign a second project in a country until we have engineers and spare modules there. In Australia we employ 38 people, two-thirds of them hired locally. Every export customer can have its operating data hosted locally and can switch off our remote access altogether. We do not sell into the United States, because the tariff and policy risk is too great for a company of our size to carry across a fifteen-year warranty. Availability across our export fleet has stayed above 98 per cent, the figure lenders ask for first when we tender again.