Some of the most consequential logistics dependencies in the world are almost invisible. A container of steel is visible. A shipment of automobiles is visible. A tanker carrying oil is visible. Their scale makes their importance obvious.

Germanium is different.

The amount embedded in an individual system can be measured in grams. Yet germanium is used in technologies where losing access to a tiny quantity of material can prevent a much more valuable product from being manufactured at all. That is exactly the kind of dependency traditional logistics metrics tend to underestimate.

The war in Ukraine is now providing a particularly dramatic demonstration. Recent estimates cited by BMO Capital Markets suggest that roughly 15 million drones deployed in the conflict during 2026 could consume approximately 15 metric tons of germanium if an average system contains around one gram. Against estimated global demand of roughly 343 tons, that would represent more than 4 percent of annual demand from an application that barely existed at this scale several years ago.

The number is interesting. The logistics lesson is more important.

A Gram That Enables the System

Germanium is a brittle, grayish-white semiconductor metal with a peculiar combination of optical and electrical properties. Those properties make it valuable in fiber-optic communications, infrared optics, semiconductor applications, specialized solar cells and radiation detectors. In infrared systems, germanium can transmit wavelengths that ordinary glass cannot, making it useful in thermal imaging and night-vision equipment.

That makes germanium relevant well beyond drones. It sits inside telecommunications infrastructure, sensors, defense systems, advanced electronics and portions of the space economy. The problem is that the amount of material required for any individual system can be deceptively small.

Traditional materials management tends to focus naturally on large flows: steel, aluminum, plastics, batteries, packaging, fuel. These consume enormous volumes, occupy warehouse capacity and generate obvious transportation requirements. But volume and operational criticality are not the same thing. One gram of material can have almost no impact on transportation cost and an enormous impact on production continuity. If a manufacturer has 99.9 percent of the material required to build a sensor but lacks the tiny quantity of germanium required for an infrared component, it does not have 99.9 percent of a finished product.

It has zero finished products.

The Byproduct Problem

Germanium presents another logistics complication: production cannot necessarily respond quickly to higher prices. It is principally recovered as a byproduct of zinc processing, with additional recovery possible from certain coal ashes and secondary sources. That means germanium output is partly constrained by the economics and processing infrastructure of another material. A sharp increase in germanium demand does not automatically produce a corresponding increase in mined germanium supply. This creates a different kind of supply elasticity problem.

If demand suddenly rises for a conventional commodity with substantial dedicated production capacity, higher prices can eventually stimulate additional production. For a byproduct mineral, the response may depend on whether sufficient host material is being mined, whether processors possess recovery capability, whether refining capacity exists and whether the economics justify extracting relatively small concentrations. The logistics chain is therefore longer than the material itself suggests.

Germanium does not simply move from mine to factory. It can depend on zinc production, concentrate flows, smelting economics, recovery technology, refining capacity, geopolitical access and the facilities capable of producing the specialized form required by the final application.

And concentration matters. USGS reports that China remained the leading global producer and exporter of germanium metal in 2025. China introduced export licensing in 2023 and subsequently banned exports of germanium to the United States in December 2024, while reported Chinese exports declined sharply. This turns an obscure material dependency into a geopolitical one.

The Logistics Network Is a Graph

Most companies still understand their materials network primarily through tiers. Tier-one supplier. Tier-two supplier. Tier-three supplier. That structure is useful, but it can hide the real dependency.

A better representation increasingly looks like a graph: Germanium -> processor -> infrared component -> sensor -> finished system -> manufacturing facility -> distribution network -> customer. Now add alternate processors, substitute components, national boundaries, export controls, transportation lanes, inventory buffers and production lead times. The real question is no longer, “Who supplies us with germanium?” It is, “Which finished products, customers and operational commitments become impossible if this node disappears?”

That is a much more important question.

AI and modern graph technologies should eventually make this type of multi-hop dependency analysis far more practical. Instead of building static supplier maps, logistics organizations can model how materials, components, facilities, regulations and customers interact and then calculate where small upstream disruptions create disproportionately large downstream consequences. The most dangerous node in a network may not be the largest. It may be the node with no substitute.

Resilience Begins With Engineering

The response to these dependencies cannot simply be “hold more inventory.” Inventory is one resilience mechanism, but critical-material exposure ultimately requires engineering choices as well.

Can the component be redesigned to use less germanium? Can another optical material substitute? Can germanium be recovered through recycling? Can alternative refining capacity be qualified? Can specifications be changed without degrading system performance? Can products be redesigned around components with more geographically diverse inputs?

These are engineering questions, but they are also logistics questions because product architecture determines logistics architecture. A product designed around a single highly concentrated material creates one type of logistics network. A product designed around interchangeable inputs creates another. The Department of Defense has already demonstrated part of this logic through efforts to recycle germanium lenses from decommissioned equipment, while U.S. industrial initiatives are attempting to expand domestic recovery and refining capacity. Those efforts reflect a broader realization: resilience can be engineered upstream rather than purchased entirely through inventory downstream.

Tiny Flows, Large Consequences

The germanium story is not really about germanium. It is about a class of dependencies that will become more important as products become more technologically sophisticated.

Advanced manufacturing increasingly depends on small quantities of highly specialized materials. Semiconductors, sensors, robotics, communications equipment, batteries, aerospace systems and defense platforms contain materials whose physical volume bears little relationship to their economic importance. Logistics organizations therefore need a different way to rank risk.

Annual spend is not enough. Shipment volume is not enough. Supplier count is not enough. The more useful question is what happens to the system when the flow disappears.

A component that costs $20 but stops a $200,000 machine deserves more attention than its purchasing value suggests. A material measured in kilograms that controls billions of dollars of downstream production deserves more attention than its freight volume suggests.

That is the emerging critical-materials problem.

The importance of a flow is not determined by its volume. It is determined by what stops moving when that flow disappears.

It may weigh one gram.

The post Drone Warfare Is Exposing a New Critical-Minerals Logistics Problem appeared first on Logistics Viewpoints.

Tesla’s Cybercab is being treated as an autonomous-vehicle story. That misses what may become the more interesting problem. Building a vehicle that can drive itself is one challenge. Building a national operating network that can finance, charge, maintain, clean, stage, reposition, and continuously utilize hundreds of thousands of those vehicles is another.

Tesla is beginning to acknowledge that distinction. The company is soliciting interest from potential partners in “Cybercab fleet vehicle purchasing” and “mobility hubs and infrastructure.” The implication is important: Tesla may not intend to own every vehicle, charging location, maintenance facility, and piece of supporting infrastructure required to build a national Robotaxi network. That moves Cybercab squarely into the world of logistics.

The Vehicle Is Only Part of the System

Autonomous driving concentrates attention on what happens inside the vehicle. At scale, however, the economics will increasingly be determined by what happens around it. A Cybercab waiting to charge is not generating revenue. Neither is one sitting in the wrong part of a city, waiting for cleaning, undergoing maintenance, or parked because demand has temporarily disappeared.

Once autonomous fleets grow large enough, Tesla will face many of the same questions transportation and logistics companies have dealt with for decades. Where should assets be positioned? How much capacity is required? How quickly can assets be turned? Where are the bottlenecks, and how much idle time can the economics tolerate?

Removing the driver changes the cost structure dramatically, but it does not eliminate fleet economics. In some ways, it makes asset utilization even more important because a larger share of the business case shifts toward maximizing the productive use of the physical asset.

Tesla Could Separate the Platform From the Assets

There is another potentially significant piece of Tesla’s approach. The company could own the intelligence layer while other businesses own portions of the physical network. Tesla would manufacture the Cybercab, provide the autonomous-driving technology, control the Robotaxi application, assign rides, establish operating standards, and coordinate the network. Independent operators could provide capital, purchase vehicles, operate local fleets, and potentially develop charging and mobility hubs.

The comparison with Amazon’s Delivery Service Partner model is not exact, but the architecture is familiar. Amazon built an enormous last-mile delivery network without directly owning every vehicle or employing every driver. It controls much of the demand, technology, process, and orchestration while independent businesses provide substantial amounts of physical operating capacity.

Tesla could potentially do something similar with autonomous mobility. That would allow the company to scale the network using outside capital while retaining control of the platform. For Tesla, that could be extraordinarily powerful. For the entrepreneurs supplying the capital, it could be considerably more complicated.

Utilization Will Determine the Economics

Assume an entrepreneur eventually buys 20 or 30 Cybercabs. The business case will not be determined primarily by how impressive the autonomous-driving system is. It will depend on vehicle acquisition costs, financing, insurance, electricity, maintenance, cleaning, facility costs, downtime, Tesla’s share of revenue, and—above all—utilization.

Transportation companies know this equation well. A truck can contain exceptional technology and still be a bad asset if it spends too much time sitting. The same will be true for autonomous vehicles. Cybercab could remove one of the largest operating expenses in passenger transportation—the driver—but the remaining asset still has to earn enough revenue across enough hours of the day to generate an acceptable return on invested capital.

That makes orchestration central to the business model. Tesla will need to predict where demand will develop, position vehicles before that demand arrives, balance charging requirements against passenger demand, schedule maintenance with minimal disruption, and continuously redistribute capacity across the network. At that point, this starts looking less like a taxi operation and more like a highly automated transportation network.

Mobility Hubs Become Logistics Facilities

Tesla’s reference to “mobility hubs and infrastructure” may ultimately be as important as Cybercab itself. A large autonomous fleet needs somewhere to go. Vehicles need charging, cleaning, inspection, maintenance, tire service, repairs, and occasionally temporary storage. They will also need to be staged near predicted demand.

At sufficient scale, these mobility hubs effectively become fleet terminals. Hundreds of vehicles could arrive and depart throughout the day. Charging capacity must be allocated, maintenance must be prioritized, and vehicles needed near peak-demand areas may have to be returned to service ahead of others. Electricity prices could even influence when charging occurs.

The hub therefore becomes another node in the network that must be optimized. This is exactly where artificial intelligence begins moving beyond the vehicle. ARC’s AI in the Supply Chain research describes a future based on connected intelligence: systems that sense changing conditions, reason across constraints, coordinate decisions, and increasingly act across interconnected operational environments.

Cybercab could become a very visible example of that architecture. The AI is not simply driving the car. Eventually, AI may be determining which car should move, where it should move, whether it should take a passenger or charge, when maintenance should occur, and how thousands of vehicles should be balanced across a metropolitan network. That is a logistics control problem.

The Network May Matter More Than the Car

This is where Cybercab becomes strategically interesting. Tesla may sell vehicles, but the network coordinating those vehicles could ultimately be much more valuable. The platform knows passenger demand, controls dispatching, and knows vehicle location, battery state, expected trip duration, charging availability, and potentially maintenance condition. As the system grows, every trip generates additional information that can improve future decisions.

More vehicles increase coverage. Better coverage attracts more passengers. More trips create better operating data, and better operating data improves positioning and utilization. Improved utilization, in turn, makes owning additional vehicles more attractive.

That is the network effect Tesla wants. If outside operators finance much of the vehicle fleet and supporting infrastructure, Tesla could potentially accelerate that effect without financing the entire physical network itself. The strategic prize would no longer be simply an autonomous car. It would be a transportation operating system sitting above a distributed fleet of physical assets.

But Someone Still Owns the Risk

There is an obvious problem for prospective Cybercab operators: Tesla would likely control many of the variables determining their economics. The platform could influence pricing, revenue sharing, dispatch priority, operating requirements, software charges, and potentially the balance between Tesla-owned and independently owned vehicles.

That is also familiar territory in logistics. Independent contractors and small transportation companies frequently operate inside networks controlled by much larger platforms. The platform brings demand and infrastructure, but it also establishes many of the rules.

Before Cybercab becomes a meaningful entrepreneurial opportunity, operators will need to understand the actual economics: vehicle price, revenue split, utilization expectations, financing, insurance, maintenance responsibility, charging costs, and contractual protections. Until then, this remains an interesting architecture rather than a proven business model.

The Bigger Logistics Story

Tesla still has significant hurdles ahead. Cybercab deployments remain limited, production must scale, regulatory questions remain, and the economics of the partner model have not yet been disclosed. But the direction is worth watching because autonomous transportation is often described too narrowly as the elimination of the driver.

The larger change may be the emergence of transportation networks in which software increasingly separates intelligence from physical asset ownership. Central platforms could control demand, routing, optimization, and customer interaction while distributed operators supply capital and physical capacity. That architecture already exists in pieces across logistics.

Tesla may be preparing to apply it to autonomous mobility at enormous scale. If it succeeds, Cybercab will be more than a self-driving taxi. It will be a logistics network that happens to move people.

The post Tesla’s Cybercab Is Really a Logistics Network appeared first on Logistics Viewpoints.

Stop managing freight costs reactively. Here’s how leading procurement teams stay ahead.

Fuel surcharges swing 15–25% within a single quarter. Annual tender cycles can’t keep pace. And yet most procurement teams still rely on manual processes, spreadsheets, and outdated rate data to manage some of their biggest cost lines.

Join our live webinar where we’ll show you exactly how agile procurement works in practice — from standardizing and automating your tendering process to managing fuel and BAF updates between tender cycles without the overhead of a full renegotiation.

What You’ll Learn in 20 Minutes:

Automate Your Procurement Workflows: See how Freightos Procure automates up to 90% of manual freight procurement tasks — from RfQ generation and carrier management to sanity checks and best-rate calculations — so your team focuses on decisions, not admin.

Manage Fuel and BAF Updates Between Tenders: Learn how to run structured BAF review cycles between full tenders, using either your own methodology or carrier-submitted updates — with built-in approval workflows and full audit trails.

Benchmark Rates Against the Real Market: See how Freightos Terminal gives you daily-updated spot and contract benchmarks sourced from $50B+ in real commercial freight spend — so you always know where your rates stand before you negotiate.

Plus, Judah Levine, Head of Research at Freightos will share what the latest market signals mean for your lanes right now.

Have questions? Bring them to the session for a live Q&A.

If you’re busy that day, register anyway and we’ll send you the full recording after.

The post Agile Freight Procurement in Practice: From Rate Management to BAF Automation appeared first on Freightos.