Last updated: August 2026
Logistics 4.0 is the set of technologies that automate the warehouse, transport and last mile in the chain that gets a product from producer to end consumer. Behind a parcel arriving within 24 hours sits a machine built from millions of robots, predictive algorithms, drones and, increasingly, humanoids still in testing: the most automated part of the digital economy, and also the least visible to the person who simply finds the parcel at their door.
Contents
- What logistics 4.0 is
- A parcel’s 24-hour journey
- Robotic warehouses: inside the black box
- Humanoid robots in the warehouse
- Dark stores and quick commerce
- Drones and rovers: automating the last mile
- Predictive AI: forecasting what you’ll buy
- Ports, ships and rail: heavy-duty automation
- Resilient supply chains: chips, tariffs and reshoring
- Italy: freight hubs and PNRR logistics funding
- FAQ and the parcel of 2035
What logistics 4.0 is (definition)
Logistics 4.0 is the application of robotics, artificial intelligence and connectivity to three layers of the distribution chain: the warehouse (storage and order picking), transport (moving goods between hubs, ports and sorting centres) and the last mile (final delivery to the customer). It’s the layer of the digital economy consumers see least of, since their only touchpoint is the moment a courier rings the doorbell, but it’s also the layer that has seen the most aggressive automation over the past decade.
A parcel’s 24-hour journey
An online order completes a journey that, in a modern facility, takes just a few hours: after the customer clicks, the order is automatically routed to the nearest warehouse with stock; a goods-to-person system brings the mobile shelf to the picker, human or robotic; a sorting algorithm routes the parcel to the correct outbound lane based on destination; a transport vehicle, often already moving on a route predicted by AI before the order was even placed, carries it to the local hub; finally, the last mile, the most expensive and least efficient leg of the whole chain, delivers it to the door. It’s the last mile, not the warehouse, that typically represents the single highest cost line in the entire delivery process.
Robotic warehouses: inside the black box
Amazon passed one million active robots in its warehouses in 2025, a number closing in on the company’s roughly 1.56 million employees: today around 75% of Amazon shipments are supported in some way by automation, and packages handled per employee have risen from 175 in 2015 to nearly 3,870. Average staff per warehouse, meanwhile, fell from 1,000 in 2020 to 670 in 2024, the lowest level in sixteen years, a figure worth including to describe the automation story without downplaying its employment impact.
| Technology | How it works | Example |
| Goods-to-person (AMR) | Mobile robots bring the shelf to the picker, instead of the other way round | Amazon Robotics (formerly Kiva Systems), Proteus |
| Grid storage | Robots move across an elevated grid above stacked bins, retrieving the order from above | Ocado, AutoStore |
| Automated sorting | Arms and belts route parcels to the correct outbound lane based on destination | Amazon Sequoia, Cardinal |
| Fleet-coordination AI | Optimises the routes of hundreds of robots in real time to cut idle time | Amazon DeepFleet (a claimed 10% cut in travel time) |
Humanoid robots in the warehouse
The warehouse is the first genuine commercial market for humanoid robots, well ahead of home applications: unlike a house, a warehouse has standardised aisles, regular shelving and repetitive tasks, ideal conditions for a humanoid still early on its learning curve. Several manufacturers, from Boston Dynamics’ Atlas to Agility Robotics’ Digit and Figure, are testing their humanoids precisely in warehouse settings, often in partnership with major retailers. Unitree, among the most price-aggressive Chinese manufacturers, is also increasingly targeting logistics applications beyond demo footage. Every new pilot announced inside a real distribution centre, not a studio demo, is a concrete signal of how close this technology is getting to deployment at scale.
Dark stores and quick commerce
The ten-minute grocery delivery economy has been through a bubble and a brutal deflation. Gorillas, the Berlin startup that popularised the model, was absorbed by Getir in 2022 and shut down entirely in 2024; Getir itself, after pulling out of Germany, the UK, the Netherlands and the US in 2024, sold what was left of its delivery business to Uber for $335 million in early 2026, a fraction of the $12 billion valuation it had reached at its peak, a roughly 97% collapse in value within a few years.
The survivors followed a different logic from debt-funded growth: Flink, in Germany, remains competitive precisely thanks to close partnerships with supermarket chains such as REWE and Carrefour, which are also its shareholders, rather than building an independent dark-store network from scratch. The lesson for the sector is that order density per urban warehouse, not the delivery speed promised in marketing, is what decides a dark store’s economic survival.
Drones and rovers: automating the last mile
The real state of drone delivery, as of August 2026, deserves an honest telling: it remains a niche phenomenon at global scale, far more mature in specific applications than in generalised urban delivery. Zipline, which specialises in delivering medical supplies to rural or hard-to-reach areas, has passed 1.4 million autonomous deliveries: it’s the use case that works best, largely because it often operates where the alternative, a road vehicle, is slower or entirely absent, rather than competing head-on with conventional urban delivery. Manna Aero, an Irish startup, has completed over 200,000 autonomous deliveries in partnership with Deliveroo in suburban Dublin, but still under favourable conditions and at limited scale.
On the European regulatory side, EASA distinguishes three risk categories for drone flight: the Open category, low risk, requires no prior authorisation but mandates visual line of sight (VLOS) flight and a maximum altitude cap; the Specific category, medium risk, allows flight beyond the pilot’s visual line of sight (BVLOS), necessary for any commercial delivery over distance, but requires a specific operational authorisation from the national authority, ENAC in Italy; the Certified category, for the highest risk (including passenger transport), is still under regulatory development, with rules expected between 2026 and 2027. In Italy, in short, large-scale commercial delivery is held back more by the complexity of Specific-category authorisation than by any technological limit of the drones themselves.
Predictive AI: forecasting what you’ll buy
Demand forecasting is the prediction, through statistical and machine-learning models, of how much of a given product will sell in a given area over a given period, cross-referencing order history, seasonality, local events and even weather forecasts. Its most aggressive evolution is anticipatory shipping: a product is moved to a hub near the customer before the order is even placed, based on the statistical probability that the purchase will happen soon. It’s one of the harder predictive-AI applications to explain to the public, because it flips the intuitive order of things: shipping doesn’t follow the order, it bets ahead of it.
Ports, ships and rail: heavy-duty automation
Automation isn’t limited to the last mile: major container ports, from Rotterdam to Singapore, already use autonomous cranes and yard vehicles to move containers between ship and stacking area. On the maritime side, several Northern European pilot projects (particularly in Norway) are testing fully autonomous short-range cargo ships on fixed, controlled routes. In Europe, cross-border rail freight corridors remain a key piece of heavy transport’s decarbonisation strategy, though technical standardisation between different member states’ networks remains a slower practical obstacle than the technology itself.
Resilient supply chains: chips, tariffs and reshoring
2026 is a year of significant instability for electronic components. In February 2026 the US Supreme Court ruled tariffs imposed under the IEEPA statute unconstitutional; new temporary tariffs of 10-15% arrived through a different legal mechanism (Section 122), while 25% tariffs on AI-bound semiconductors remain in place. For the Italian electronics industry, which had grown its exports to the US by 12% in the first half of 2025, planning costs in this environment has become harder: according to trade association ANIE, 56% of companies in the sector still report concrete supply-chain issues.
Reshoring, bringing production back to one’s own country, is often cited as the logical response to these tensions, but in Italy it runs into a constraint more underrated than cost: a shortage of skilled technicians available on the labour market. Companies are handling the volatility with three complementary strategies: diversifying suppliers, nearshoring to nearby countries, and building targeted strategic stockpiles for the highest-risk components, from logic ICs to memory chips.
Italy: freight hubs and PNRR logistics funding
Italy’s network of interporti, the intermodal hubs where rail and road freight meet, is the least-told part of the national logistics story. The Quadrante Europa interporto in Verona remains Italy’s first and Europe’s second in importance. Bologna is a reference hub for the Centre-North, while the Nola interporto, in Southern Italy, closed 2025 with 50% growth in rail traffic (2,000 trains handled) and is targeting over 3,000 trains a year in 2026, helped by the arrival of international logistics operators such as Kuehne+Nagel. Italy’s Ministry of Infrastructure and Transport has allocated PNRR recovery funds to digitalise sixteen interporti of national relevance, aligned with the interoperability standards of the National Digital Logistics Platform, with the stated goal of bringing the Italian network up to the European standard of freight trains up to 750 metres long, against a current national average of around 550 metres.
FAQ
Will drones deliver parcels in Italy any time soon?
Not at scale in the near term: the Specific category required for beyond-visual-line-of-sight (BVLOS) flight requires a specific operational authorisation from ENAC, an obstacle that is more regulatory than technological. Healthcare applications in remote areas remain the most mature use case globally.
How many robots does Amazon really have in its warehouses?
Over one million, a number closing in on the company’s roughly 1.56 million employees; around 75% of Amazon shipments are today supported in some way by automation.
Why did ten-minute grocery delivery services fail?
Because the model required very high order density per urban warehouse to be economically sustainable; those who didn’t reach that density, such as Getir and Gorillas, burned capital at an unsustainable pace, while those who partnered with existing supermarket chains, such as Flink, survived.
Will humanoid robots replace warehouse workers?
Too early to say for certain: even in the most advanced settings they remain in the pilot stage today, but the warehouse, thanks to its repetitive, standardised structure, remains the commercial environment most favourable to their adoption at scale.
Does the 2026 chip shortage affect consumer products too?
Yes, indirectly: tensions over AI-bound semiconductors are absorbing production capacity and investment attention, with knock-on effects on delivery times and prices for other categories of electronic components.
The parcel of 2035
If today’s trends keep maturing, the parcel of the next decade could travel along an almost entirely automated chain: warehouses populated by humanoid robots alongside today’s AMRs, drones deployed beyond today’s healthcare niches once the BVLOS regulatory knot is resolved, and predictive AI able to anticipate purchases with ever-narrower margins of error. The open question is how much human labour will remain in a supply chain that, from Amazon downward, is already showing how quickly the number of people needed to move the same volume of goods can shrink.
Sources
- Amazon, corporate data on active warehouse robots and automation (2025)
- Zipline, Manna Aero, corporate data on cumulative autonomous deliveries
- EASA, ENAC, regulatory framework for Open, Specific and Certified flight categories
- Agenda Digitale, analysis of quick commerce and Gorillas’ exit from Italy
- Business Model Innovation, account of Getir’s sale to Uber (early 2026)
- ANIE (Italian federation of electrotechnical and electronic companies), 2026 supply-chain data
- Italian Ministry of Infrastructure and Transport, PNRR funds for interporti digitalisation
- Interporto di Nola, Interporto Quadrante Europa Verona, Interporto Bologna, corporate data and statements
