Across freight, cold storage and circular electronics markets, several large platforms such as Full Truck Alliance, ATRenew and Lineage are choosing a different way to defend utilisation: improving match quality instead of cutting price.
In Brief
- ATRenew, Lineage and Full Truck Alliance are improving utilisation by raising match quality rather than cutting price, tightening who can transact and how capacity is allocated.
- AI dispatch, automated inspection and stricter platform rules are steering the right load, pallet or device to the right node, lifting fulfilment rates and protecting margins.
- The trade-off is slower headline growth and stricter participation rules as operators prioritise cleaner data, stronger guarantees and more reliable network economics.
The Underlying Pattern and Stakes
The common thread across China’s AI-enabled freight and circular electronics platforms, and a global cold-chain operator, is a deliberate decision to improve match quality and utilisation before reaching for price as the primary lever. Whether the asset is a truck, a pallet of frozen seafood or a used smartphone, the focus is shifting from filling capacity at any cost to curating what enters the network, how it is priced and how reliably it flows.
The shift is not about generic digitisation or AI experimentation. Instead, ATRenew, Full Truck Alliance and Lineage are redesigning networks, contracts and control layers so that algorithms, guarantees and governance determine who can transact, on what terms and with what level of protection, while list prices largely hold or even rise.
How Companies Are Converging
The first area of convergence is the decision to use governance to raise the floor on transaction quality, even when that slows headline growth. Full Truck Alliance spent the back half of 2025 stripping noncompliant and low-value activity out of its freight platform: enforcing real name verification on both shippers and truckers, cracking down on freight reselling, and reclassifying mislabelled carpool loads that had been masquerading as full-truck orders. The result was slower order growth in the fourth quarter, but the overall fulfillment rate climbed to around 42.7 percent, with mid- and low-frequency direct shippers approaching 65 percent. ATRenew is doing something similar on the resale side: standardised inspection and pricing tools for franchise partners, compliant refurbishment, and category-specific take rates that penalise poor quality supply while rewarding higher-standard goods.
In parallel, these operators are tightening the economic logic of who gets what capacity. Lineage has leaned into minimum storage guarantees and volume commitments rather than chasing occupancy through discounting. Minimum guarantees rose to roughly 46.7 percent of capacity, up 180 basis points sequentially, even as the U.S. cold storage market sat with an estimated 9.5 percent excess capacity. Full Truck Alliance pushed commission penetration on fulfilled freight to almost 89 percent across 273 cities, and ATRenew continued to refine marketplace and consignment take rates by category, with over-6 percent B2B take rates for merchants on PJT and high single digits in the Paipai consignment model.
The second point of convergence is the deeper use of AI and automation to steer and vet flows, not to undercut price. ATRenew’s network of more than 2,000 AHS stores and nearly 300-city coverage is underpinned by automated inspection in recycling centres and AI for service and training. That infrastructure allows compliant refurbished revenue to grow more than 100 percent year on year and lifts 1P gross margin from 11.7 to 13.4 percent, while keeping fulfilment cost as a share of revenue slightly down despite higher volumes. Full Truck Alliance is embedding an AI assistant directly into shipper workflows: a shipper can describe a load via voice and the agent handles listing, trucker screening, negotiation and matching, raising both the number of valid trucker bids and the proportion of fully automated transactions. Lineage’s proprietary LinOS execution layer is delivering double-digit gains in units per hour at pilot sites, holding same-warehouse labour spend flat on a 1.5 billion dollar labour base even as throughput slips slightly.
A third shared move is the choice to use platform and marketplace logic to manage match quality across highly fragmented participants. ATRenew’s PJT marketplace aggregates more than 1.37 million small and medium merchants, while Paipai’s consignment service takes over inventory management, traffic and aftersales for smaller sellers in exchange for higher take. Full Truck Alliance is pushing more of its freight into structured commission-based flows, with shippers and carriers assessed via scorecards that account for cancellation behaviour, complaint rates and service quality. Lineage’s equivalent is more contract-based: new developments are largely customer-led, with strict volume and revenue guarantees and less speculative capacity, meaning that who gets storage and at what commitment level is decided up front, not at the dock door.
Operating Model Mechanics
Underneath this pattern sit very concrete decisions about nodes, contracts and control layers.
ATRenew’s model hinges on a dense hybrid network: self-operated AHS stores in top-tier cities, joint-operated locations in lower tiers, and nearly two thousand to-door agents. Those nodes feed C2B trade-ins from partners such as JD.com and device brands into refurbishment hubs where automated inspection systems grade condition and route devices either into high-margin 1P-to-consumer channels or into B2B flows on PJT. Multi-category recycling, now live in 878 self-operated and 131 franchise locations, is layered on top to push more high-value items through the same physical nodes, with new stores stabilising within two to three months and generating roughly 7,000 renminbi of monthly contribution each.
Full Truck Alliance’s network is less physical and more algorithmic, but the mechanics are similar. Shippers and truckers are pulled into the platform through strict real name and vehicle verification, and every interaction runs through in-app messaging and protected voice channels. Commission-based orders are orchestrated by scoring models that weigh route reliability, historic cancellations and dispute behaviour. The AI assistant becomes a dispatch layer that sits on top of this governance spine, selecting a subset of truckers to engage, testing pricing and locking in a match that is more likely to complete without incident. As this layer matures, a larger share of loads bypass manual brokers and ad hoc phone calls, but within a much narrower band of acceptable behaviours.
Lineage is in the business of hard capacity, not platforms, yet it is applying a similar logic through contracts and system rules. In a market where U.S. public refrigerated capacity grew around 14.5 percent over four years while relevant food volumes rose only 5 percent, the company has resisted the temptation to trade price for volume. Instead, it has re-based volume guarantees after the pandemic destocking phase and is now landing new contracts and customer-led developments with guarantees above the portfolio average. Minimum storage commitments, together with LinOS-driven labour and energy gains, mean that same-warehouse revenue per occupied pallet can edge up by about 1 percent even as physical occupancy hovers in the mid-70s.
In practice, this kind of configuration typically requires a few common elements:
- A master data and identity layer that controls who can transact and what attributes drive matching and pricing.
- Contract and rule engines capable of enforcing guarantees, scorecard cut-offs and category-specific take rates inside day-to-day execution.
- A feedback loop from fulfillment outcomes back into these rules so that cancellations, dwell times and service failures directly influence future access and economics.
Where AI shows up, it is mostly in making these layers responsive and scalable. ATRenew’s automated inspection and AI-powered customer service compress cycle times and reduce subjective grading errors, feeding more consistent quality data into pricing and category decisions. Full Truck Alliance’s dispatch agent takes historically messy negotiation and matching processes and pins them to structured data points that can be optimised over time. LinOS pushes real-time labour and task data into management dashboards that show productivity at the pallet level, allowing local teams to adjust processes without ripping out rate cards.
Risk, Constraints and Trade-offs
This operating stance carries real constraints. Cleaning up a marketplace or tightening access rules often hits the top line before it helps the bottom line. Full Truck Alliance’s ecosystem governance removed fake accounts and freight reselling that previously inflated order counts with little monetisation; fourth quarter order growth slowed to just over 12 percent even as commission revenue grew almost 30 percent. ATRenew temporarily reduced take rates in gold recycling to protect user economics as commodity prices spiked, sacrificing some short-term service revenue to maintain trust and growth in a standardised, low-margin category.
Holding price while chasing better utilisation also pushes more risk onto planning and capital allocation. Lineage is carrying a network built through 70 acquisitions over five years and 25 buildings in ramp-up at a time when throughput is flat to slightly down and tariffs have cut West Coast container volumes by around 20 percent. Idling eight underutilised facilities and consolidating into adjacent nodes lowers labour and energy cost, but it also concentrates operational risk into the remaining buildings and assumes that new guarantees will hold when trade flows normalise.
There is also the tension between algorithmic control and local nuance. ATRenew’s inspection systems and pricing tools standardise how a phone, luxury bag or gold item is valued, but category economics vary: luxury can carry service take rates above 10 percent, gold only low single digits. Applying a single rule set across all categories would either leave money on the table or overburden low-margin flows. Full Truck Alliance’s AI dispatch models might favour the most reliable truckers based on scores, but over-concentration on a small subset could weaken coverage and responsiveness in less dense corridors.
Finally, the decision to use guarantees and commitments as the primary utilisation tool commits future capacity. Lineage’s higher minimum storage guarantees and Full Truck Alliance’s deepening share of commission-based loads both improve visibility and economic occupancy, yet they reduce flexibility to reallocate space or trucks if demand shifts abruptly across regions or sectors. ATRenew’s push to move more small merchants into consignment effectively centralises inventory risk; operational failures in a single hub or quality issue can ripple across hundreds of sellers who now depend on the same processing node.
Operational Self-check
The pattern described here is hard to fake. It shows up in how nodes, contracts and control layers behave under stress, not in slogans about AI or platforms.
Questions that expose whether this pattern is present or absent include:
- Does match quality improve when growth slows, or do cancellations, complaints and discounting spike as networks try to fill capacity?
- Are guarantees, take rates and scorecards actively managed by category and corridor, or are they set once and left untouched despite clear changes in mix and behaviour?
- Do AI and automation initiatives sit inside the execution layers that shape who can transact and where flows go, or are they mostly peripheral pilots that do not alter capacity or contract decisions?
What This Pattern Signals
Across these examples a structural shift is visible: capacity-heavy and marketplace-based operators are choosing to manage freight, storage and secondary device flows through higher-quality matches, stronger guarantees and more disciplined participation rules, rather than relying on price cuts to defend utilisation. AI is being used to enforce and operationalise those rules at scale, not to create a separate layer of experimentation.
If this approach persists, network and sourcing design will increasingly revolve around where governance and control are strongest, not simply where capacity is cheapest. Nodes that can enforce guarantees, run automated grading or host AI dispatch will become the preferred anchors, and contracts will be written to steer more volume through those locations, even if that means idling or repurposing other sites.
This looks less like a temporary crisis response and more like the next phase of maturation for asset-intensive and platform-heavy supply chains in China and beyond. As oversupplied markets, tariff volatility and circular flows become normal, operators that can tune match quality and utilisation without eroding price will be structurally better placed than those that rely on blunt volume tactics.
This article is based on recent earnings reports and public disclosures from the companies referenced.
