The most alluring vision of autonomous driving is being stress-tested in the most rugged and gritty of environments: open-pit mines.

EACON (HKEX: 07687) officially commenced its public offering on the Hong Kong Stock Exchange on June 29th. This company does not operate in the familiar urban Robotaxi space, nor is it a consumer internet darling. Its focus is a narrower, more capital-intensive, and production-oriented business: autonomous driving solutions for mining operations.

Shifting Focus to the Mining Pit

Moving the focus from capital markets to the open-pit mines makes the story more concrete. Dust, gravel, slopes, loading zones, and dumping sites stretch out before you, with massive hundred-ton haul trucks shuttling back and forth. For drivers, this isn't ordinary "driving"; it's more like enduring a shift amidst machine noise, dust, and high-intensity repetitive labor. For mining companies, it's a stark financial reality: labor is hard to find and expensive, safety responsibilities are immense, and equipment must run nearly continuously to maintain hauling capacity.

It is precisely for these reasons that while autonomous driving is often pitched as a future story for city streets, the mining industry has emerged as a crucible for its commercial viability. Here, there are no complex interactions with public traffic or pedestrians, but there are closed environments, fixed routes, clear demands, harsh conditions, heavy-load transport, and continuous operation. Clients are willing to pay for safety and efficiency, but this environment is also enough to deter many players who only excel at single-vehicle demonstrations.

EACON has grown in this niche. It is neither a vehicle manufacturer nor a pure software company. Instead, it focuses on the haulage segment of mining, providing autonomous haul trucks and related solutions. According to its prospectus, EACON ranked first in China's mining autonomous driving solutions market in 2025.

This Hong Kong IPO means the capital market must re-evaluate an autonomous driving business that "goes underground." Its appeal isn't based on futuristic roadshow pitches, but on whether its systems can work consistently in the dirtiest, most grueling, and most dangerous places.

Sifting Out "Pseudo-Autonomous" Systems On-Site

A common misconception is that mines are closed, simple environments, making them easier than city roads. The closed nature is an advantage, reducing the unpredictability of open traffic, but closed does not mean simple. The difficulty in mines is of a different, industrial-system variety.

City roads test capabilities in mixed traffic, traffic lights, and rules. Mines test continuous operation on unpaved roads where grades and conditions change with excavation progress. Dust, glare, rain, snow, and low light affect perception. Vehicles must also coordinate with excavators, bulldozers, and manned vehicles during loading, hauling, and dumping cycles.

Especially in the heaviest and most hazardous haulage part of open-pit mining, a haul truck's job isn't simply going from point A to B. Taking overburden removal as an example, the truck must haul waste rock to the dump site, stop precisely in front of the dump wall, and complete the tipping process. Stopping too far means material misses the target area; getting the rear wheels too close to the wall risks a slide-off.

Therefore, the core challenge isn't getting one vehicle to run, but getting a fleet of vehicles, a dispatching system, loading/unloading equipment, and on-site personnel to operate stably within the same production schedule. What cross-over autonomous driving players most often underestimate is the gap between a "single-vehicle demo" and "large-scale mixed fleet operations."

Beyond this, mine operators face real pressures. Following a 2023 accident at an open-pit coal mine in Alxa, safety regulations have tightened, and policies are pushing for mine automation and reduced personnel. According to relevant requirements, by 2026, intelligent production capacity should account for no less than 60% of national coal output, and intelligent equipment or robots should replace human labor in no less than 30% of hazardous/heavy positions in coal mines and 20% in non-coal mines.

Coupled with an aging driver workforce, recruitment difficulties, and intense shift work, autonomous mining haulage isn't just a tech concept. It's an industrial demand squeezed out by the combined pressures of safety, regulation, labor, and production efficiency. This explains why mining companies aren't buying hype. What they truly care about is fewer accidents, less downtime, fewer hires, and whether haulage targets per shift can be consistently met.

Turning Vehicles into a System

EACON's core product, "Zhushan," is an autonomous haul truck product and solution for closed environments. Simply put, it's not about adding an "autonomous driving function" to a truck. It's about providing a complete unmanned operation solution for the mining haulage scene, encompassing environmental perception, vehicle execution, and dispatching coordination.

This solution is driven by the company's autonomous driving technology, modular drive-by-wire chassis platform, proprietary electronic/electrical architecture, and dispatching platform. The goal is to achieve unmanned haulage in mines under standard and extreme working conditions. Its applications span coal, metal, and limestone mines. The core value is reducing reliance on human drivers, improving safety, and providing more stable haulage efficiency at scale.

EACON's competitive moat lies here as well. The prospectus shows that as of end-2025, its system is compatible with over 70 models from brands representing over 90% of the haul truck market, with a single-mine fleet exceeding 500 autonomous vehicles. Its standardized, modular architecture can compress deployment time for a new mine to weeks, with the fastest deployment taking 3 days.

Its "Yushi" chassis is like a "motherboard," adaptable to different OEMs like Tongli, Yutong, XCMG, and to different powertrains like diesel, range-extender, and pure electric.

These OEMs provide the vehicles and chassis, forming the industrial chain foundation. However, the real value-add after autonomous trucks enter the mining production line comes more from vehicle perception/decision-making, cloud dispatching, mixed-fleet coordination, data loops, and on-site operations. Especially in the unstructured environment of a mine—with no clear lane markings, roads that change with excavation, and slower braking/steering responses for heavy vehicles—the algorithmic logic differs from that of passenger car Robotaxis.

For mining autonomous solution providers, whoever can stably integrate more vehicle models, more mines, and more loading/unloading cycles gets closer to becoming the production nerve center for mining companies.

The First Commercial Hurdle: From Pilot Projects to Real Production Lines

The biggest fear for mining autonomy is getting stuck as a pilot project. Having a few vehicles running in one mine is very different from long-term, large-scale operation across multiple mines. The former is about tech demos; the latter is about production assets.

The core data EACON currently presents relates to deployment scale and customer retention. By the end of 2025, the company had deployed 2,580 active autonomous haul trucks across all its mine sites. Measured by the number of active autonomous vehicles, it held a 55.5% market share in China's mining autonomous driving solutions market.

The client side tells a similar story. As of end-2025, the company had 13 terminal client groups, with solutions deployed across 30 mines. Clients include large mining enterprises or their operating systems like China Energy Investment Group, State Power Investment Corporation, TBEA, Zijin Mining Group, Shougang Group, Guanghui Energy, and Baowu Steel Group. From 2023 to 2025, the company maintained a 100% retention rate across all terminal client groups.

In the B2B mining context, retention rate is more meaningful than in B2C internet products. Once a mine integrates a supplier's solution into its haulage system, dispatching processes, and safety management, the cost of switching is significant. A client staying typically means the system is embedded in the production workflow; subsequent expansion will also build upon the existing processes, data, and operational framework.

Public information shows that as of 2024, EACON's cumulative domestic operational mileage exceeded 40 million kilometers. By 2025, it had covered over 20 large mines in China, including 7 of the country's top 12 open-pit coal mines.

As revenue expanded, R&D expenses reached 2.7 billion yuan in 2025, with the R&D expense ratio maintained around 20%. During the same period, capital expenditures consistently exceeded depreciation and amortization, indicating the company is still investing for scale expansion and delivery networks.

Shift to Asset-Light Model, Improving Gross Margins

In the early days of mining autonomy, suppliers often had to own and operate the vehicles themselves, using heavy capital investment to lower client adoption barriers. This helped open the market but had clear drawbacks: vehicle purchase, depreciation, maintenance, and operations all fell on the supplier, with revenue growth potentially eroded by asset costs.

From the company's perspective, EACON's business model can be divided into vehicle-owning and non-vehicle-owning categories. In the vehicle-owning model, the solution provider fully purchases the autonomous trucks, providing the vehicles, operations management, and technical support. In the non-vehicle-owning model, the client owns the truck assets, while the provider supplies the necessary autonomous driving hardware/software, technical support, and subsequent operational services. From the client's view, this can be further broken down into outright purchase, subscription, and operations-based models.

Calculations from Soochow Securities show that the vehicle-owning model initially yields gross margins of only about 1.4% at project deployment, potentially rising to 26.5% after normalization. In the non-vehicle-owning model, subscription-based gross margins are around 70.8%, while an outright purchase model, annualized over a 5-year cycle, yields gross margins around 71.4%. The reason is simple: revenue from software, algorithms, and services doesn't bear the cost of vehicle purchase and heavy depreciation, leading to naturally thicker profits.

EACON's own data is validating this shift. From 2023 to 2025, the proportion of revenue from the client-provided fleet model rose from 41.7% to 46.0%, and then to 56.8%, becoming the dominant force in the revenue structure. During the same period, company revenue was 2.7 billion yuan, 9.9 billion yuan, and 14.4 billion yuan, respectively, representing a three-year compound annual growth rate of 130.2%. Gross margin improved from -18.6% to 7.6%, and then to 10.1%.

At a glance, a 10.1% consolidated gross margin isn't spectacularly high. However, this financial statement reflects a transition: early heavy-asset projects, R&D investment, on-site delivery, and scale expansion all weigh on overall margins. What's truly worth watching is whether the proportion of the non-vehicle-owning model can continue to increase, and whether software/service revenue can pull the consolidated gross margin higher.

Large Market Potential, Overseas as a Tougher Benchmark

China's autonomous mining truck market is entering an acceleration phase. Data from Frost & Sullivan shows that by the end of 2025, the number of operational autonomous haul trucks in China exceeded 4,000. Domestic penetration of autonomous truck sales rose from 1.1% in 2021 to 12.1% in 2025, and is projected to reach 52.1% by 2030. The domestic market size for autonomous mining solutions is projected to reach 22.3 billion yuan by 2030, with a CAGR of about 42.5% from 2026 to 2030.

The overseas market provides a more direct economic case. As of 2024, there were over 1,600 autonomous vehicles in overseas open-pit mines, with Caterpillar and Komatsu holding over 90% share of the overseas autonomous truck market, and Australia accounting for about 70% of the overseas distribution. According to 2024 public operational data from Caterpillar and Komatsu, applying autonomous technology in open-pit mines reduced haulage costs by 15%, increased tire life by 40%, and improved transport efficiency by 30%.

Taking the Solomon Iron Ore mine in Australia as an example, with 60 autonomous trucks operating alongside 12 manned ones, the autonomous trucks could operate up to 23.5 hours per day, with only about 0.5 hours for inspection and refueling, providing 2-3 more operational hours per day than manned trucks. Annual operating time reached nearly 7,000 hours, compared to the typical 5,500-6,000 hours for human drivers, translating to a 20-30% productivity increase. This data shows overseas mines adopt autonomy not just for being "advanced," but because the economics work.

However, overseas does not mean easy. Caterpillar and Komatsu started early, have deep client relationships, and robust service networks. Meanwhile, many early overseas solutions relied more on first-generation AGV architectures, depending heavily on rule-based cloud and vehicle path-following, with limited adaptability to complex mines. With Australia's autonomous truck penetration around 15% in 2024, blue ocean space remains. If domestic solution providers can leverage their "smart cloud + smart vehicle" architectural differences to enter, the opportunity is significant, but the barriers are high.

EACON has identified Australia as a key focus for internationalization. The company has established an office there, built cooperation with mining service systems like Thiess, and is advancing a localization pilot in Kalgoorlie with Norton Gold Fields, a subsidiary of Zijin Mining. On December 18, 2025, EACON, in collaboration with Zijin Mining and global mining services provider Thiess, completed the first "safety-driver-off" test at the Norton Gold Fields project in Australia. This was the first on-site "unmanned" test of Chinese mining autonomous driving technology completed in Australia.

If this path succeeds, the overseas market could raise the growth ceiling. Simultaneously, overseas compliance, on-site delivery, and local service networks will impose tougher requirements on the company.

The Post-IPO Story Returns to Operational Metrics

The Hong Kong IPO is not the finish line for EACON.

As China's top-ranked mining autonomous driving company, EACON has gained an early advantage through its hardware-software integration, compatibility with numerous truck OEMs, a roster of leading clients, and a 100% customer retention rate.

What's most worth watching next is whether the proportion of the non-vehicle-owning model can continue to rise, whether software and algorithm services can lift gross margins, and whether operating cash flow and losses can gradually narrow.

Mines aren't impressed by slick presentations. They only recognize one thing: whether the system can work consistently in the dirtiest, most grueling, and most dangerous places.

This is the most valuable, and also the most brutal, part of EACON's story. The application defines the technology; scale validates commercial viability. The players who truly endure must first go down into the mine before they can talk about the future.