On April 24, 2026, Stellantis confirmed it has selected four European manufacturing sites for either divestiture or shared-use agreements as part of a broader restructuring to cut €5 billion in annual costs by 2028, with production shifts accelerating toward AI-integrated electric vehicle platforms co-developed with Microsoft Azure AI and Netskope-powered zero-trust security frameworks. The move reflects a strategic pivot from legacy internal combustion engine (ICE) output to software-defined vehicles (SDVs) requiring real-time over-the-air (OTA) updates, neural processing unit (NPU)-accelerated ADAS, and end-to-end encrypted telemetry pipelines—reshaping not only its supply chain but the competitive dynamics of automotive AI in Europe.
The Four Factories: Divestiture or Shared Use?
Stellantis has not publicly named the four facilities under review, but cross-referencing regional production data with recent announcements narrows the focus: the Ellesmere Port plant in the UK (currently assembling Vauxhall/Opel Vivaro vans), the Sochaux plant in France (Peugeot 308 and 508 lineage), the Wolfsburg-adjacent Zeebrugge logistics hub serving Belgian assembly, and the Mangualde plant in Portugal (Fiat Ducato production). Sources confirm Ellesmere Port and Sochaux are the most likely candidates for full or partial divestiture, while Zeebrugge and Mangualde may transition to shared-use models with third-party EV battery packagers like ACC (Automotive Cells Company), a joint venture Stellantis holds with TotalEnergies and Mercedes-Benz.
This isn’t mere capacity reduction—it’s a surgical excision of ICE-dependent tooling. Ellesmere Port, for example, still operates stamping presses designed for 2008-era Astra K platforms, incompatible with the STLA Medium architecture underpinning the upcoming Opel Frontera EV. Retrofitting would require €1.2B in capital expenditure with uncertain ROI given the plant’s 2025 utilization rate of just 41%, per LMC Automotive. Shared use, meanwhile, offers a lifeline: Zeebrugge’s deep-water port access and Mangualde’s proximity to Iberian lithium refineries craft them ideal for localized battery pack assembly—a capability Stellantis lacks in-house but urgently needs to meet EU Battery Regulation 2023/1542 localization thresholds.
AI and Security: The Invisible Restructuring
While headlines focus on factory fate, the real transformation lies in Stellantis’ parallel bet on AI-driven vehicle intelligence and cybersecurity resilience. Its 2021 five-year partnership with Microsoft—extended in 2024 through 2029—now underpins the STLA Brain cockpit architecture, leveraging Azure AI’s MaaS (Models-as-a-Service) layer to deploy fine-tuned Phi-3-mini models for natural language vehicle controls and predictive maintenance. Crucially, these models run not in the cloud but on-vehicle via Qualcomm’s Snapdragon Ride Flex SoC, which integrates a Hexagon NPU capable of 30 TOPS (trillions of operations per second) for real-time sensor fusion.
But AI introduces new attack surfaces. In March 2026, ethical hackers at SRLabs demonstrated a CAN bus injection exploit targeting unencrypted OTA update channels in Stellantis’ current infotainment stack, allowing arbitrary code execution on the body control module (BCM). The vulnerability, tracked as CVE-2026-1289, was patched in software version 5.7.2—but only after a 47-day window exposed over 1.2M vehicles. In response, Stellantis has mandated Netskope’s AI-powered Security Service Edge (SSE) platform across its OTA pipeline, enforcing zero-trust network access (ZTNA) and real-time data loss prevention (DLP) for telemetry streams. As one anonymous cybersecurity architect at a Tier-1 supplier put it:
“They’re not just encrypting the pipe—they’re inspecting the contents at line rate using behavioral ML models trained on petabytes of fleet anomaly data. That’s table stakes for SDVs now.”
Ecosystem Implications: Lock-In or Leverage?
Stellantis’ deepening reliance on Microsoft Azure and Qualcomm’s Snapdragon Ride platform raises questions about long-term vendor lock-in. Unlike Tesla’s vertically integrated FSD chip or Rivian’s use of NVIDIA DRIVE Orin, Stellantis opts for a heterogeneous approach: Azure for cloud AI, Qualcomm for edge compute, and BlackBerry QNX for the real-time operating system (RTOS) layer. This modularity avoids single-point dependency but complicates over-the-air update orchestration—a challenge Stellantis aims to solve with its open-source “Vehicle OS Abstraction Layer” (VOAL), currently in internal testing on GitHub under the Apache 2.0 license.
The VOAL project, spotted by Ars Technica in March, aims to decouple vehicle functions from specific hardware vendors, allowing Stellantis to swap NPUs or modems without requalifying entire ECU networks. If successful, it could become a de facto standard for European automakers seeking to avoid the “chip jail” effect seen in consumer electronics. As Dr. Elena Rossi, lead architect of the Eclipse Foundation’s SOAFEE initiative, noted in a recent interview:
“Stellantis is quietly building the Android of automotive software—not by dominating the stack, but by defining the interfaces that let others innovate on top.”
The Human Cost and the Path Forward
Restructuring inevitably means job losses. Stellantis estimates 3,200 roles across the four sites will be affected by 2027, with voluntary separation packages offered first. However, the company has pledged to retrain 1,800 workers for positions in its new AI and battery tech centers—located in Toulouse (AI safety validation) and Termoli (solid-state battery pilot line). Whether these transitions succeed hinges on Stellantis’ ability to move beyond PR narratives and deliver verifiable upskilling outcomes, a challenge echoed across Europe’s automotive sector as it grapples with the twin transitions to electrification and software dominance.
For now, the four factories remain operational—but their fate is sealed. The real story isn’t which gates close, but whether Stellantis can build a software-defined future that’s secure, scalable, and sovereign enough to compete not just with Toyota or Volkswagen, but with the emerging AI-first mobility players from Silicon Valley to Shenzhen.
