At its 629 (June 29) conference in Anji, Chinese forklift vendor Zhongli (中力) pushed embodied loading robots and 3 m/s transfer units, all vendor stated. The question is whether any holds up outside the demo hall.
On June 29 in Anji, Zhejiang, Zhongli (中力), one of China's largest industrial-truck and material-handling vendors, opened the eighth edition of its 629 Innovation Conference with a troupe of its own robots dancing the Butterfly Lovers violin concerto alongside human performers, a visual the vendor cast as a "from can-perform to can-deploy" metaphor. Across the two days that followed, Zhongli pressed a larger argument: embodied AI, the use of AI-powered robots that load trucks, run yards, stock warehouses, and serve ports, is rewriting how Chinese industry moves goods, and the category itself will be redrawn within five years.
What the dance was selling, the product line, raises the harder question. Zhongli framed 629 around a set of "five qualitative changes" it says the material-handling industry is undergoing: from devices to scenes, from carriers to objects, from handling to mobility, from in-yard to full-domain, and from autonomous to intelligent-autonomous control (Leiphone). On the loading floor the headline was a 2.0-ton embodied loading robot the vendor says can empty a standard wing-door truck in about 60 seconds with a stated cargo-gap tolerance of 2 cm or less, supported by a 1.5-ton covered-van unit, a 2.0-ton side-shift variant, and a transfer cart called the VLM801 (Leiphone; Sohu). Inside the warehouse the lineup ran from an XLM101 smart stacker and an XPL201 high-speed handler that Zhongli says runs near 3 m/s, to an XCD061 ultra-low-profile fork robot, an XSL201i dual-lift handler, and a 1.0-ton omnidirectional unit (Leiphone).
Every one of those figures is Zhongli's own. The 60-second cycle, the 2 cm gap tolerance, the near 3 m/s transfer speed, the "embodied brain" architecture that pairs multimodal perception with autonomous decision-making, and the "five-year restructuring" call all came from Zhongli-controlled settings (Leiphone; Sohu). The "five qualitative changes" framework is positioning rather than measurement. The published recap does not include independent third-party benchmarks on cycle time, uptime, payload variety, or deployed fleet size for these specific units.
Through the first half of 2026, Chinese industrial-robotics vendors have collectively turned embodied AI from a research phrase into a procurement phrase, with embodied loading, AI-driven truck yards, and full-scene mobility becoming standard pitch frameworks (TOM News). Each vendor paints a different silhouette of the same shift. Independent third-party data on what any of them are actually running at scale is harder to find than the roadmaps.
The unit-level numbers hide as much as they show. The 60-second cycle depends on lighting, weather, mixed-SKU pallets, racking tolerance, and what happens after the robot finishes. The 2 cm cargo-gap tolerance only matters if real warehouses enforce that standard, and most do not. A 3 m/s transfer speed requires a receiving side that can keep that pace and safety systems that survive a year of mixed human-and-robot traffic.
Zhongli is not a startup pitching a single demo. Capital-markets analysis of the public company treats it as a "shovel-seller" in the embodied-robotics transition, the supplier that benefits regardless of which downstream architecture wins (Eastmoney). The 629 line is already eight editions deep: a 2014 launch of small-format lithium forklifts, a 2020 oil-to-electric transition, and the 2026 embodied-AI edition. The customer base behind those numbers is real, and the company has shipped enough electric industrial trucks to make the embodied claims harder to dismiss as stagecraft.
On the technical side, the showcase holds up. Multimodal perception paired with autonomous decision-making across outdoor and indoor transitions is consistent with the 2026 class of embodied systems, and the breadth of the lineup suggests engineering depth rather than a single demo unit (Sohu). The narrative packaging is where the showcase overreaches. The "five qualitative changes" framework and the "five-year restructuring" line are useful narratives, and the vendor's own recap makes that distinction explicit (Leiphone). The buyer's question in late 2026 is whether any of these units are running at meaningful uptime outside the demo hall. Zhongli has not yet answered that publicly.
The next test is not the ninth 629 conference in 2027. It is whether independent industry analysts, Tier-1 logistics operators, and Zhongli's own periodic disclosures let outsiders track what is deployed, where, on what terms, and at what uptime. Until then the stage demo and the loading floor will keep telling different stories, and Zhongli will keep asking the industry to trust the first one.