One of the exercises we do at Kvistlund at the start of each year is to map the European supply chain software landscape — which layers are well-served, which are underinvested, and where we see the most interesting white space for new companies. This is our working view going into 2025. It's informed by the 13 investments we've made since 2021 and by the deal flow we see, which gives us a reasonably current picture of where founders are building and where capital is flowing.
The framing I find most useful is to think of the supply chain software stack as a series of layers, each of which processes information from the layer below it and produces decisions or instructions for the layer below. This is admittedly a simplification — real supply chains are messier and more networked than a clean stack diagram suggests — but it's a useful simplification for thinking about where the investment opportunity concentrates.
The physical layer: hardware and connectivity
At the base of the stack is physical infrastructure: warehouse equipment, vehicles, carriers, and the connectivity that links them. This layer is increasingly instrumented — IoT sensors on conveyors, GPS on vehicles, RFID on pallets, telematics on trucks — and that instrumentation produces the raw data that the layers above it consume. The investment opportunity in the physical layer itself is mostly hardware and infrastructure, which is outside our thesis. What's interesting to us is the software that translates physical-layer data into operational intelligence.
One observation worth making: the physical instrumentation of European supply chains is materially less complete than in the US. Telematics penetration in European road freight is lower. Warehouse automation penetration — measured as percentage of square meters with automated picking or sortation systems — is lower in European mid-market warehouses than in comparable US operations. Cold chain monitoring consistency is variable across the continent. This instrumentation gap is simultaneously a problem for the software layers above (less data to work with) and an opportunity (more headroom for improvement once the data is there).
The execution layer: TMS and WMS
The execution layer — transportation management systems and warehouse management systems — is the most capital-intensive layer in European supply chain software and the hardest to displace. The major European TMS and WMS vendors have been selling highly configured enterprise implementations for twenty years, and the switching costs are real. An operator who has spent 18 months implementing a WMS and has trained 200 warehouse staff on its interface is not looking to change systems without a compelling reason.
We do not typically invest at the execution layer. The displacement cycle is too long, the integration complexity is too high for a seed-stage company, and the incumbents have strong enough positions to delay competitive displacement for years. The interesting investments we've made adjacent to the execution layer — companies like Kargo in warehouse intelligence and FourKites in supply chain visibility — are building on top of existing execution systems rather than displacing them. The layer above WMS and TMS, not the systems themselves, is where the value creation opportunity is currently most accessible for early-stage companies.
The intelligence layer: where we invest
Above the execution layer sits what I call the intelligence layer: software that takes operational data from TMS, WMS, and physical infrastructure, applies ML or mathematical optimization, and produces better decisions than the execution layer would make on its own. This is where Kvistlund's investments are concentrated. Predictive logistics (Portcast), freight visibility with ML-enhanced ETAs (Logixboard), fleet intelligence (Optimal Dynamics), decision science infrastructure (Nextmv), supply chain visibility (FourKites), warehouse intelligence (Kargo), waste stream analytics (Greyparrot), digital freight operations (Flexport) — these are all intelligence layer plays.
The intelligence layer in Europe is materially undercapitalized relative to the US, which is the core investment thesis. A freight forwarder running a modern TMS in Germany has access to less AI-enhanced intelligence tooling than a comparable US freight forwarder, partly because fewer European-native intelligence companies have reached the scale and maturity needed to displace US software options, and partly because the US investor base has historically funded more of these companies at earlier stages.
The coordination layer: where the white space is
The layer that I find most interesting — and most underinvested in Europe — sits above the intelligence layer: supply chain coordination software. This is the software that connects the intelligence outputs from multiple systems and multiple parties into coherent, cross-network decisions.
Here's a concrete example of the problem. A shipper's demand forecasting model predicts a volume spike in week 10. Their inventory positioning system responds by increasing safety stock at distribution centers. Their carrier selection model identifies the optimal carriers for the expected volumes. Each of these intelligence systems is doing its job. But they are not communicating with each other or with the shipper's carrier network. The carrier network doesn't know about the demand spike until the shipper actually tenders the loads. The carriers could have adjusted their own capacity allocation if they'd had two weeks of visibility. The coordination failure wastes capacity on both sides — the shipper ends up paying spot rates for capacity that the carriers would have allocated at contract rates if they'd known earlier.
Supply chain coordination platforms — multiparty networks where demand signals, capacity signals, and risk signals are shared upstream and downstream through the supply chain with appropriate privacy controls — are significantly less developed in Europe than in Asia. The Asian export corridor has several well-funded coordination platforms with real multiparty adoption. The European B2B supply chain has not yet produced a comparable network. The opportunity is real, but it requires solving harder problems: data sharing agreements between competitors, privacy and antitrust compliance in a GDPR environment, and the multi-sided network bootstrapping problem. None of these are insurmountable, but they are the reasons the category hasn't moved faster.
Where the investment is overconcentrated
Being honest about where we think the market is getting crowded: freight visibility is overinvested relative to the available differentiation. There are now dozens of companies offering "real-time freight tracking" in Europe, many of which are building on the same underlying carrier data sources and differentiating primarily on UX rather than on the quality or uniqueness of the intelligence they provide. The freight visibility category is going to consolidate. Companies that have built genuine data network advantages — proprietary carrier integrations, carrier-contributed data that doesn't flow to competitors — will survive consolidation. Companies that are reselling the same data as everyone else with a different interface will not.
Last-mile optimization in dense urban environments is also getting crowded. The algorithmic and ML approaches are not sufficiently differentiated in that segment to support a large number of independent companies at scale. The last-mile optimization companies that have survived the early consolidation wave are the ones that found a specific customer segment — urban grocery delivery, pharmaceutical last mile, returns-intensive apparel — where their product has genuine operational depth rather than generic route optimization capability. Generic last-mile optimization is increasingly a commodity feature of TMS platforms, not a standalone product.
The white space, as we see it, is in the coordination layer, in procurement intelligence, in cold chain software (which we wrote about separately), and in the physical world instrumentation layer where computer vision and IoT are creating data signals that didn't exist before. That's where we're spending most of our time in 2025.