A container is waiting at a port. The vessel carrying it has been delayed, the customer needs the goods within four days, and the warehouse that would normally receive them is almost full. A truck is available, but its battery is scheduled to recharge during the cheapest electricity window of the day. Another route is faster, although it passes through a region where severe weather is expected.
Today, resolving this problem might involve a freight forwarder, a transport planner, a warehouse manager, a fleet dispatcher and several software systems. Each sees part of the situation. Someone has to assemble the picture, weigh the alternatives and make a decision.
The next generation of logistics technology is being built around a different proposition: what if the network could assemble that picture itself?
Not merely display the delay, but calculate its consequences. Not merely recommend another truck, but reserve it. Not merely identify a charging opportunity, but incorporate the electricity price into the transport decision. Not merely report that a storm is approaching, but estimate the financial exposure and determine whether the risk should be avoided, retained or insured.
This is the direction in which supply-chain technology is moving. It is also where the industry becomes much more consequential than the familiar story about robots replacing warehouse workers or autonomous trucks replacing drivers.
The emerging contest is over who controls the intelligence that coordinates the physical economy.
That intelligence will sit above an enormous installed base of ports, warehouses, vehicles, ships, factories and energy infrastructure. It will draw on conventional optimisation, artificial intelligence, digital twins and, potentially, quantum computing. It will increasingly incorporate risk models and insurance. And it will operate in an energy system where electricity can sometimes be extraordinarily cheap, but where access to power at the right place and time may become a valuable competitive advantage.
China is approaching this transition from a particularly strong position in manufacturing, batteries and integrated digital-commerce networks. The United States has important advantages in AI infrastructure, enterprise software and autonomous highway freight. Europe combines powerful logistics incumbents, industrial automation, truck manufacturing and regulation. Other regions will matter through ports, energy, manufacturing, trade corridors and specialised operating environments.
The next decade will not produce one winner. It will produce a new distribution of power across the system.
From software that records decisions to software that makes them
The first important change is already under way.
Traditional logistics software was largely designed to record and coordinate transactions. A warehouse management system knew where inventory was stored. A transportation management system planned shipments. A telematics platform reported vehicle location. An enterprise system recorded orders and invoices.
These systems were valuable, but they usually depended on people to interpret the information and decide what to do next.
Agentic AI changes the ambition.
An AI agent is not simply a chatbot attached to a database. In an operational setting, it is software that can pursue a defined objective, use tools, retrieve information, evaluate alternatives and take authorised actions. Several agents may work together, with one responsible for inventory, another for transport, another for customer commitments and another for energy or risk.
The distinction matters because logistics is full of decisions that are individually small but collectively expensive.
Should a shipment be consolidated with another load? Is it worth paying for faster transport? Should inventory be moved before a forecast demand increase? Can a delivery be delayed until electricity is cheaper? Which customer should receive scarce stock when supply is interrupted?
The economic value lies in making these decisions consistently across a network rather than optimising each department separately.
China's JD Logistics provides one of the more concrete examples of how far this approach is progressing. In its 2025 ESG report, the company described its Super Brain LLM 2.0 as an agentic decision system connected to a digital twin of people, vehicles, goods and facilities. It reported that the system could reduce certain complex planning problems involving tens of millions of variables from days to under two hours and coordinate decisions across warehousing, transportation and last-mile delivery. These are company-reported results, not independently established industry benchmarks, but they show the direction of investment.
The same report disclosed RMB 4.1 billion of research and development spending in 2025 and an R&D workforce of 4,902. More than 20 of its LangzuTech automated warehouses were operating by the end of that year.
The significance is not that every warehouse will soon run itself. It is that a major logistics operator is trying to connect prediction, planning and physical execution within one architecture.
That is a much more difficult undertaking than building an AI assistant.
It requires reliable data, operational permissions, integration with existing systems, safety controls and a clear understanding of what happens when the model is wrong.
A mistaken answer in a chatbot is inconvenient. A mistaken instruction to dispatch a truck, release inventory or alter a charging schedule can create a financial or safety problem.
For that reason, the most credible near-term model is not unrestricted autonomy. It is bounded autonomy: agents act within defined limits, routine decisions are automated, and consequential exceptions are escalated to people.
Over time, those boundaries may expand as systems demonstrate reliability.
Why China is building more than electric trucks
China's position in this transition is often described through electric vehicles. That understates the broader industrial strategy.
The country has developed a dense ecosystem connecting battery manufacturing, commercial vehicles, heavy industry, logistics platforms, e-commerce, charging infrastructure and increasingly sophisticated software.
The IEA estimates that electric trucks accounted for about one-quarter of Chinese truck sales in 2025. China produced more than 90% of the world's electric trucks, while CATL supplied approximately 80% of the batteries used in Chinese electric trucks.
Those figures describe more than a successful vehicle market. They describe an industrial system in which the principal components, manufacturers and customers are located close enough to reinforce one another.
The next stage is to connect that system to energy infrastructure.
In May 2025, CATL announced its Qiji standardised heavy-truck battery-swapping programme and a proposed national network of major freight corridors. Its stated ambition was to cover 80% of China's trunk transportation capacity by 2030. That is a corporate target, not an achieved outcome, but it reveals the scale of the strategy.
Battery swapping is particularly interesting for commercial vehicles because the economics differ from passenger cars.
A truck is a revenue-generating asset. Time spent charging is time unavailable for freight movement. If a standardised battery can be exchanged quickly, the operator may be able to separate vehicle utilisation from battery charging.
The battery can charge when electricity is cheap. The truck can continue working.
That creates a new business model in which the battery, the vehicle and the energy service need not all be owned by the same party.
It also creates a new source of industrial power. A company controlling a widely adopted battery standard and swapping network could influence vehicle compatibility, energy procurement, financing and fleet operating economics.
CATL's more recent integrated charging-and-swapping architecture extends this logic. The company says its passenger-vehicle and heavy-truck swapping stations will incorporate high-power charging, allowing the same infrastructure to serve multiple replenishment needs. Its reported efficiency and utilisation improvements remain company claims, but the strategic direction is clear: the charging station is becoming an energy-management platform rather than a simple electricity dispenser.
China's digital-commerce companies are working on the other side of the same problem.
Cainiao describes a supply-chain technology suite connecting order management, warehousing, transportation, settlement, forecasting and control-tower functions. Its network also gives it access to large volumes of cross-border logistics data.
Alibaba.com introduced an agentic AI sourcing mode in late 2025 designed to compare suppliers across pricing, logistics, certifications and production capabilities. The commercial importance is that procurement intelligence may increasingly connect directly to fulfilment intelligence.
If a platform can help a buyer select a supplier, arrange production, finance the transaction, book freight, manage customs and coordinate delivery, it occupies a much more powerful position than a conventional online marketplace.
It becomes part of the operating infrastructure of trade.
That is one of the most important Chinese technology strategies to watch over the next decade.
The global contest is over different layers of the system
China's strength does not mean it will dominate every part of future logistics.
The United States has a different collection of advantages. Its AI and cloud ecosystem, enterprise software industry, large domestic freight market and autonomous-driving companies provide a strong foundation for software-defined logistics.
Aurora began regular driverless commercial freight deliveries between Dallas and Houston in May 2025. That established an important operational milestone, although it does not establish that autonomous trucking is ready for unrestricted deployment. https://ir.aurora.tech/
Daimler Truck, through its majority-owned autonomous-driving company Torc Robotics, is pursuing a different but complementary strategy: integrating Level-4 autonomy into a truck platform designed with the necessary redundant systems. Torc has targeted commercial hub-to-hub operations in the United States in 2027.
Europe's strengths lie elsewhere. It has globally competitive truck manufacturers, major logistics groups, industrial automation companies and a regulatory framework that is actively pushing heavy transport toward lower emissions.
The EU's heavy-duty vehicle CO₂ standards require substantial reductions in average emissions from covered new vehicles, including a 45% reduction for 2030–2034 relative to the relevant 2019 baseline. That creates a regulatory demand signal for manufacturers and infrastructure investors. EU heavy-duty vehicle CO₂ regulation
Meanwhile, DSV, DHL and Kuehne+Nagel possess something that new AI companies cannot easily manufacture: established customer relationships, global operating networks and large quantities of real operational data.
Their opportunity is to turn those networks into intelligent platforms.
Their risk is that software companies or customers themselves capture more of the decision-making layer.
The question is not whether logistics incumbents will use AI. They already are.
It is whether they will own the orchestration layer or become execution providers beneath somebody else's software.
Quantum computing: a potentially important tool, not a magic solution
Quantum computing belongs in this outlook, but it requires more careful treatment than it usually receives.
Logistics contains many difficult optimisation problems. A fleet operator may need to assign hundreds of vehicles to thousands of deliveries while respecting capacity, time windows, driver rules, charging constraints and customer priorities. A warehouse may need to sequence thousands of tasks while avoiding congestion. A manufacturer may need to schedule production across machines, components and delivery commitments.
These problems can become computationally difficult as the number of possible combinations grows.
Quantum computing is being investigated as one way to solve certain classes of optimisation problems. Quantum-inspired algorithms, which run on conventional computers but borrow ideas from quantum methods, are also being developed.
The distinction is important.
A quantum-inspired algorithm is not evidence that a quantum computer has achieved a practical advantage.
And a successful quantum demonstration is not proof that the technology will outperform the best classical methods across real logistics networks.
IBM has explored hybrid classical-quantum approaches to vehicle routing, including a study involving delivery to 1,200 locations in New York City with time-window and capacity constraints. The work illustrates the potential application, but it should not be interpreted as proof of general commercial quantum superiority.
A more concrete industrial example comes from D-Wave and Ford Otosan. In March 2025, the companies announced that a hybrid-quantum application had entered production to improve vehicle manufacturing sequencing. D-Wave reported that scheduling 1,000 vehicles fell from approximately 30 minutes to less than five minutes. This is a company-reported result in a specific manufacturing application, not a universal logistics benchmark.
The broader research remains cautious. A systematic review published in December 2025 found promise in quantum and quantum-inspired transport optimisation but identified hardware scale, realistic data, benchmarking and total computation time as important limitations.
That suggests a sensible 5–10 year thesis.
Quantum computing is unlikely to replace conventional logistics software wholesale. If it becomes commercially useful, it will probably appear as a specialised optimisation service inside a larger classical and AI system.
An agent might identify a difficult scheduling problem, formulate it, send it to a suitable optimisation engine and then evaluate the result against operational constraints.
The user may never know whether the underlying solver was classical, quantum-inspired or quantum.
What matters is whether it produces a better decision at an acceptable cost and speed.
China is also investing in the underlying quantum-computing ecosystem. Origin Quantum, for example, describes its Origin Pilot operating system as a platform for coordinating quantum hardware with high-performance computing, AI and classical workloads. This demonstrates work on hybrid computing infrastructure, but it does not establish that China has achieved a commercial quantum advantage in logistics.
The opportunity is real enough to monitor. The evidence is not yet strong enough to build a logistics investment thesis around quantum advantage alone.
The rise of algorithmic risk management
The next important change may be less visible than autonomous vehicles.
It concerns risk.
Traditional supply-chain risk management often begins with a list of suppliers, locations and possible disruptions. Companies identify vulnerabilities, purchase insurance, hold inventory and create contingency plans.
The difficulty is that modern supply chains are networks of dependencies.
A factory may depend on a component supplier that depends on another supplier in a different country. A port closure may affect several manufacturers simultaneously. A flood may interrupt a transport route without damaging the goods themselves. A cyberattack may stop a warehouse even though the building and inventory remain physically intact.
The financial consequences propagate through the network.
Swiss Re has developed research with UC Berkeley on modelling business-interruption risk propagation through complex supply chains. Its approach involves mapping multiple supplier tiers, identifying critical nodes and estimating how disruptions affect production volumes and financial losses.
That points toward a more sophisticated form of risk management.
Instead of asking only, "What could go wrong?", companies can increasingly ask:
What happens to revenue if this supplier fails?
Which alternative route preserves the most value?
How much inventory is economically justified?
Which risks should be retained?
Which should be transferred to an insurer?
How much would a faster recovery be worth?
This is where AI, digital twins and insurance begin to intersect.
An agentic system could eventually monitor the network, identify a rising exposure, simulate alternatives and recommend a response. The response might involve moving inventory, changing suppliers, rerouting freight or purchasing additional risk protection.
The important point is that risk management becomes part of operational decision-making rather than a separate annual exercise.
Algorithmic insurance could become part of the logistics operating system
Insurance is also changing.
The relevant concept is usually called parametric insurance, rather than simply algorithmic insurance.
Traditional insurance generally pays according to an assessed loss covered by the policy. Parametric insurance pays when a predefined, measurable event occurs, subject to the policy's terms.
For example, a company might purchase cover linked to a specified river-water level, wind speed or rainfall threshold. If the agreed trigger is met, the payout is determined according to the contract without requiring the same conventional loss-adjustment process.
Munich Re explicitly offers parametric weather solutions for supply-chain, logistics and transportation risks, including disruption caused by low river levels.
This is particularly relevant because some logistics losses are difficult to insure through conventional physical-damage policies.
A shipment may not be damaged, but a closed waterway can still create substantial additional transport costs. A supplier may be unable to operate because of a weather event, causing a downstream manufacturer to lose production.
Parametric cover can provide faster liquidity in such circumstances.
But it has a limitation that should not be ignored: basis risk.
The trigger may occur without the insured suffering the expected loss, or the insured may suffer a loss without the trigger being met.
That means better data and better modelling are essential.
The future opportunity is not simply automated claims.
It is the integration of risk modelling, operational decisions and financial protection.
A logistics platform might eventually estimate that a particular route has an elevated disruption probability, calculate the expected financial exposure and compare three choices: accept the risk, reroute the shipment or purchase a suitable insurance product.
That would make insurance part of the economics of dispatch.
It would also create difficult questions around transparency, data ownership, pricing fairness and regulatory oversight.
The insurer must understand the risk it is accepting. The customer must understand what is covered. And the algorithm must not create the illusion that uncertainty has disappeared merely because it has been assigned a number.
The economics of near-zero-cost energy
The phrase "zero-cost energy" is attractive, but it needs precision.
Solar and wind generation have very low marginal operating costs once the infrastructure is built. Under certain market conditions, wholesale electricity prices can fall to zero or become negative when supply exceeds demand and the system lacks sufficient flexibility.
That does not mean electricity is free.
Generation assets require capital. Networks require investment. Storage costs money. Grid connections can be expensive. Retail tariffs include charges that do not disappear simply because wholesale prices are low.
The distinction matters enormously for fleets.
The IEA reported that negative wholesale prices accounted for approximately one-quarter of hours in South Australia during both 2023 and 2024. It also documented increasing negative-price periods in several European markets. These episodes reflect system conditions, not a permanent supply of free electricity.
For a fleet operator, however, even intermittent cheap electricity can be valuable.
A diesel truck generally purchases fuel at a price determined by the fuel market and taxes.
An electric fleet may have more flexibility.
Vehicles can charge overnight. Some can charge during periods of abundant solar generation. Stationary batteries can store electricity. Charging can be scheduled around delivery requirements and electricity tariffs.
The economic opportunity is to turn flexibility into lower operating cost.
Imagine a distribution centre with solar generation, a stationary battery and a fleet of electric trucks. The software knows which vehicles must leave at 5 a.m., which can remain parked until midday and which have enough charge for their next assignment.
It can then allocate energy according to operational need and price.
That is not free energy.
It is intelligent energy procurement and utilisation.
And it could become a meaningful competitive advantage.
A company with access to cheap renewable electricity, sufficient grid capacity and flexible charging may operate electric fleets at a lower cost than a competitor purchasing power at less favourable times.
The advantage could be especially important in high-utilisation freight operations, where energy expenditure is a substantial component of total cost.
China's battery-swapping strategy is relevant here because it separates the charging schedule from the vehicle's working schedule. Australia is also an interesting market because abundant renewable generation can create periods of low wholesale prices, although the economics depend on location, network access, tariffs and storage.
The next generation of fleet software may therefore optimise not only kilometres and delivery windows, but also electricity.
Energy abundance could change where logistics infrastructure is built
If electricity becomes cheaper in some locations and more expensive or constrained in others, logistics geography may begin to change.
Historically, distribution centres have been located according to land cost, labour availability, road access and proximity to customers.
Those factors will remain important.
But large electric fleets and automated warehouses introduce another consideration: power.
A warehouse containing extensive robotics, refrigeration, automated sorting and a large charging depot may require substantial electrical capacity.
A location with cheap land but a weak grid connection may become less attractive than one with better energy infrastructure.
Ports face similar questions as shore power, electric cargo-handling equipment and alternative-fuel infrastructure develop.
This creates a new intersection between logistics, real estate and energy investment.
The valuable asset may not simply be a warehouse.
It may be a warehouse with a large, reliable grid connection, on-site generation, storage and the ability to support a high-utilisation electric fleet.
That is a different kind of industrial property.
And it may become increasingly difficult to replicate in locations where grid capacity is scarce.
The supply chain becomes a financial system as well as a physical one
There is a deeper connection between all these developments.
Agentic AI improves the ability to coordinate decisions.
Quantum and advanced optimisation may improve the ability to solve difficult planning problems.
Digital twins improve the ability to model the physical network.
Parametric insurance improves the ability to transfer certain measurable risks.
Cheap renewable electricity improves the economics of some physical operations.
Together, they point toward a supply chain in which physical, informational and financial decisions become increasingly integrated.
A shipment is no longer merely a box moving from one place to another.
It is an asset with a location, condition, delivery commitment, energy requirement, risk exposure and financial value.
The system coordinating it may eventually make decisions across all of those dimensions.
That is the real meaning of agentic orchestration in logistics.
Not a collection of chatbots.
A decision architecture connecting the physical economy.
What could happen by 2031
The next five years are likely to be characterised by selective deployment rather than universal transformation.
Agentic AI should become more common in logistics planning, customer service, procurement and exception management. The strongest applications will be those connected to reliable operational data and clearly defined actions.
Warehouse automation will continue expanding, particularly in high-volume facilities where repetitive tasks and utilisation justify the investment.
Electric trucks will gain share in suitable duty cycles, especially in China and in European markets where regulation and infrastructure support adoption.
Fleet-energy management will become more important as operators learn to coordinate charging, electricity procurement and vehicle utilisation.
Autonomous trucking may expand across selected highway corridors, but the economics and safety performance of commercial deployment will determine the pace.
Parametric insurance and advanced supply-chain risk analytics should become more relevant for companies exposed to climate, transport and supplier disruptions.
Quantum optimisation will remain an area of experimentation and selective commercial application. It may become useful in particular scheduling or routing problems, but broad quantum advantage should not be assumed.
The common thread is that these technologies will be adopted where they solve a measurable economic problem.
What could happen by 2036
Ten years out, a more integrated system becomes plausible.
A large logistics operator could have AI agents coordinating inventory, transport, warehousing and energy across a network.
Digital twins could simulate disruptions before operational decisions are made.
Specialised optimisation engines could solve difficult scheduling problems, potentially including quantum or quantum-inspired methods where they demonstrate an advantage.
Electric fleets could charge according to both freight demand and electricity-market conditions.
Autonomous trucks could operate on selected high-volume corridors.
Risk models could continuously estimate exposure across suppliers, routes and facilities.
Insurance products could become more closely connected to measurable operational risks.
But none of this requires the entire supply chain to become autonomous.
Human judgement will remain important where decisions involve safety, commercial relationships, legal responsibility, unusual events and competing objectives.
The more plausible future is a system in which people supervise increasingly capable decision-making infrastructure.
What could change the view
The most important uncertainty is whether these technologies can be integrated economically.
AI may become much more capable, but fragmented data could prevent it from making reliable operational decisions.
Quantum hardware may improve, but classical optimisation may continue advancing fast enough to limit the commercial advantage of quantum approaches.
Electric trucks may become cheaper, but grid connections and charging infrastructure could slow adoption.
Renewable electricity may become abundant, but network constraints and tariffs could prevent fleet operators from accessing it cheaply.
Parametric insurance may expand, but basis risk, regulation and the availability of reliable data could limit its usefulness.
China may retain its manufacturing advantage, but trade restrictions, local-content requirements and the need for overseas service networks could constrain international expansion.
Autonomous trucking may demonstrate attractive economics, or safety incidents and regulatory changes could slow deployment.
And geopolitical fragmentation could make global supply chains more regional, reducing some efficiencies while increasing demand for resilience and coordination.
These are the developments that should determine whether the thesis strengthens or weakens.
Where the real opportunity lies
The most interesting opportunities may emerge between industries that have historically operated separately.
Logistics companies will need energy expertise.
Energy companies will need to understand fleet operations.
Insurers will need better supply-chain data.
Software companies will need to understand physical constraints.
Warehouse operators will need robotics and systems-integration capabilities.
Fleet managers will need to understand batteries, charging and electricity markets.
Risk professionals will need to understand network modelling and data science.
And AI specialists will need to understand why a theoretically optimal decision may be operationally impossible.
The valuable capability is not simply knowing one technology.
It is understanding how several technologies interact inside a real business.
That is also where future careers are likely to emerge: at the intersection of operations, engineering, software, energy and risk.
The larger conclusion
The future supply chain is not merely a faster version of the present one.
It is becoming a system in which decisions about goods, vehicles, energy, infrastructure and financial risk are increasingly connected.
China has built an unusually strong position in the physical foundations of that system through batteries, electric vehicles, manufacturing scale and integrated logistics networks. The United States has important advantages in AI, software and autonomous freight. Europe has powerful industrial incumbents and regulatory leverage. Other regions will compete through energy, infrastructure, trade corridors and specialised capabilities.
The next decade will test whether these advantages can be combined into systems that deliver better economics.
That is the central question.
Not whether AI can plan a route.
Not whether a quantum computer can solve an optimisation problem.
Not whether an electric truck can travel a particular distance.
But whether all of these capabilities can be coordinated to move goods more reliably, at lower total cost, with less energy and better management of risk.
The companies that solve that problem will not simply participate in logistics.
They will help determine how the physical economy operates.