🕶️ A Dutch startup claims it's 100x better than Nvidia

Nvidia's moat is narrowing, not collapsing , but European startups are betting that inference, not training, is where sovereign compute finally finds margin. Meanwhile, Loop just raised $95M to prove that in applied AI, defensibility looks less like owning models and more like owning messy, operationally critical data at insane resolution.

If they're right, the real leverage isn't in the GPU; it's in the harness that turns logistics exhaust into prescriptive action before containers seize up.

2026 will redefine IT as a strategic driver of global growth. Automation, AI-driven support, unified platforms, and zero-trust security are becoming standard, especially for distributed teams. This toolkit helps IT and HR leaders assess readiness, define goals, and build a scalable, audit-ready IT strategy for the year ahead. Learn what’s changing and how to prepare.

Download the Toolkit

Supply chains used to be about clipboards; now they’re about GPUs. Loop just raises $95M to scale its supply-chain AI, in a Series C led by Valor, the same firm backing Musk’s xAI. That cap table is a bet that whoever tames logistics’ ugliest data will own the margin stack.

Loop builds prescriptive logistics AI on top of PDFs, emails, ERP feeds, TMS data, and supplier exhaust, coordinating in-house and frontier models into one “harness.” Today it finds money leaks fast; the real play is predicting disruptions and prescribing actions before containers, cash flow, or working capital seize up.

This sits inside a broader freight automation land grab, but Loop is going deep instead of broad. If they’re right, defensibility in applied AI looks less like owning models and more like owning messy, operationally critical data at insane resolution.

In AI chips, gravity has long pulled toward Santa Clara; Europe is quietly sketching its own map. Dutch startup Euclyd is seeking nine-figure funding to build Nvidia alternatives optimized for inference, not training.

Its architecture processes data in multiple locations instead of shuttling it through a memory stack, aiming to claim 100x efficiency over Nvidia’s Vera Rubin for inference workloads. The catch: all of this is pre-scale, with real deployments not expected until late decade.

A broader European wave: Optalysys, Fractile, Arago, Axelera, Olix; rides export controls, TSMC concentration risk, and a sovereign compute agenda, pushing capital homeward.

The opportunity is real, but so are the constraints: underdeveloped foundry capacity, conservative governments, and slower deployment cycles mean the Nvidia moat is narrowing, not collapsing.