▣ DATA BRIEF · INDUSTRIAL AI
☀ Key Stats
◆ Around 80% of industrial data in the European Union remains unused, according to European Commission evidence cited by the World Economic Forum.
◆ Early trials of the Starlings energy-sharing project at the Port of Rotterdam delivered efficiency gains of around 5%, according to the report.
◆ Modelling for the Rotterdam project suggests deeper data sharing and system-wide optimization could more than triple the initial benefit, while adding shared infrastructure could increase it to as much as 8.5 times the original value.
◆ At China’s Ordos-Envision Net-Zero Industrial Park, data collaboration among more than 50 enterprises has accompanied approximately 80% on-site green electricity consumption and around 10% lower overall energy costs.
◆ Enterprise Resource Planning systems were used by 37.9% of small EU enterprises in 2023, compared with 86.3% of large enterprises, illustrating a large digital-readiness gap.
◆ About 72% of organizations in research cited by the report reported an increase in cyberattacks, while at least 35.5% of data breaches in 2024 originated from third-party incidents.
◆ The report draws on a World Economic Forum network of 42 industrial clusters across 23 countries that contribute approximately $900 billion to global GDP.
◆ Its framework was informed by more than 60 senior executives and experts through consultations and workshops.
Continue reading ↓
Around 80% of industrial data in the European Union is never used, according to European Commission evidence cited in a new World Economic Forum report on data sharing across industrial clusters.
The finding highlights a basic obstacle facing industrial artificial intelligence: companies can generate large amounts of data while still struggling to combine and use it across organizational boundaries.
The report, Intelligent Industrial Ecosystems: Turning Proximity into Intelligence, was produced by the World Economic Forum in collaboration with Accenture.
It draws on the Forum’s network of 42 industrial clusters across 23 countries, which collectively contribute approximately $900 billion to global GDP, as well as consultations with more than 60 executives and experts.
◎ HEADLINE NUMBER
80%
The share of EU industrial data that the European Commission estimated was never being used.
The unused-data figure predates the new report
The headline statistic is not a new measurement produced by the Forum in 2026.
It comes from European Commission material supporting its Data Act proposal in 2022, which identified legal, economic and technical barriers that were preventing industrial data from being reused.
The Forum argues that the challenge becomes more complicated when data must move between independent companies.
Businesses may be reluctant to share operational information because of commercial sensitivity, unclear ownership, incompatible systems, uneven digital capabilities and cybersecurity risks.
Small and large companies face a wide digital gap
One measure of that uneven readiness comes from the use of Enterprise Resource Planning systems, which integrate core business processes and data.
37.9%
Small EU enterprises using ERP systems in 2023
86.3%
Large EU enterprises using ERP systems in 2023
Eurostat figures cited in the report put ERP adoption at 37.9% among small enterprises and 86.3% among large enterprises in the EU in 2023.
The gap matters because a shared industrial data system can only work if participating companies can structure, interpret and exchange information consistently.
↳ CASE STUDY · PORT OF ROTTERDAM
Rotterdam trial delivered around 5% efficiency gains
The Port of Rotterdam’s Starlings project provides one of the report’s clearest examples of companies attempting to optimize energy use across an industrial cluster rather than separately.
The project allows participating companies to coordinate utility flows such as electricity, steam and hydrogen in near real time while limiting the amount of sensitive operational data they expose.
According to the Forum report, the first two trials delivered efficiency gains of around 5%.
The Port of Rotterdam invested approximately €2 million in the pilot, which was designed to test whether companies could coordinate energy and utility use across the cluster.
↗ FROM PILOT TO MODELLED SCALE
~5% efficiency gain in early trials
→ more than 3× the benefit in deeper optimization modelling
→ up to 8.5× with shared infrastructure
The larger numbers are not observed efficiency gains.
Modelling cited by the report suggests more advanced data sharing and system-wide optimization could more than triple the earlier benefit, while shared infrastructure could raise the modeled value to as much as 8.5 times the original level.
Important distinction: The 5% figure describes reported trial performance. The larger multipliers are modelling scenarios for deeper collaboration and additional shared infrastructure.
↳ CASE STUDY · INDUSTRIAL ENERGY
Chinese industrial park reports 10% lower energy costs
The Ordos-Envision Net-Zero Industrial Park in Inner Mongolia provides another example, this time involving data from more than 50 industrial enterprises.
A shared platform coordinates industrial facilities, wind and solar generation, energy storage and demand across participating companies.
~80%
On-site green electricity consumption
~10%
Reduction in overall energy costs
The report says the park has achieved approximately 80% on-site green electricity consumption while reducing overall energy costs by around 10%.
Those outcomes are project-level results reported by the park’s developer and do not establish that data sharing alone caused the changes.
Connecting more companies also connects their cyber risks
The same connectivity that makes cross-company optimization possible can create additional cybersecurity exposure.
Research cited by the report says 72% of organizations reported an increase in attacks, while at least 35.5% of data breaches in 2024 originated from third-party incidents.
A separate survey cited later in the report found that 78% of CEOs at highly cyber-resilient organizations identified supply-chain and third-party dependencies as their most significant cybersecurity challenge.
That makes security a system-level issue in industrial clusters: one participant’s weakness can potentially affect organizations connected to the same data, infrastructure or supply network.
The report proposes a three-stage path to data sharing
The Forum’s proposed framework starts with identifying a concrete business use case rather than building a large data platform first.
The second stage establishes the collaboration model, while the third puts applications into operation and looks for opportunities to scale them.
◇ THE REPORT’S FRAMEWORK
1 · Value case — Identify the decision, action or outcome that shared data could improve and determine whether the expected benefit justifies the cost and risk.
2 · Enablement — Establish governance, data standards, technology and the rules that determine who can access and use information.
3 · Execution — Put applications into operation, measure results and add new participants, datasets or uses when they create additional value.
Across all three stages, the report identifies four foundations: data governance, data readiness, technology and cyber resilience.
The framework is a recommended approach developed from consultations and case studies, rather than a statistical finding that guarantees better performance.
✦ Why it matters ✦
Much of the discussion around artificial intelligence focuses on what individual companies can do with their own data.
The industrial opportunity described in this report is different because energy systems, ports, supply chains and shared infrastructure often cross company boundaries.
A factory can optimize its own equipment, but it cannot independently decide when another plant should reduce steam demand, when a neighboring site has spare capacity or how several companies should respond together to grid constraints.
That broader coordination requires organizations to share enough information to make joint decisions while protecting commercially sensitive data.
The Rotterdam and Ordos cases suggest there can be measurable economic and energy benefits when that coordination works, although they do not show that the same results would occur in every industrial cluster.
The issue is becoming more relevant as electricity demand grows and AI becomes more deeply embedded in industrial systems, a trend also reflected in recent StatsJournalist coverage of global energy demand.
The challenge is that every additional connection also introduces questions about trust, competitive advantage, technical compatibility and cybersecurity.
For industrial AI, the constraint may therefore be less about whether algorithms exist and more about whether companies can create the shared data foundations required to use them across an entire system.
ⓘ How to read the findings
The headline 80% unused-data figure comes from European Commission material published in 2022. It is cited by the 2026 WEF report but is not a new measurement produced by the Forum.
The Port of Rotterdam’s approximately 5% efficiency gain describes results reported from early Starlings trials. The claim that benefits could more than triple, or reach up to 8.5 times the original value with shared infrastructure, comes from modelling rather than observed performance at that scale.
The Ordos figures of approximately 80% on-site green electricity consumption and around 10% lower energy costs are project-level outcomes reported in a case study sourced to Envision. They do not isolate how much of either outcome was caused specifically by data collaboration.
The 37.9% and 86.3% ERP adoption rates refer to small and large EU enterprises respectively in 2023. They are used as an indicator of uneven digital maturity rather than a direct measure of companies’ readiness to participate in industrial data-sharing systems.
The report’s 42 clusters across 23 countries describe the World Economic Forum’s Transitioning Industrial Clusters community. They should not be interpreted as a statistically representative sample of global industry.
The report was informed by more than 60 senior executives and experts, including insights from six in-person and virtual workshops held in 2025–2026 and one-to-one consultations. This is expert and case-study evidence rather than a global survey of industrial companies.
The cybersecurity figures come from separate sources cited by the report and use different samples and definitions. They should not be combined into a single estimate of cyber risk in industrial clusters.
The report’s three-stage framework and four enabling foundations are recommendations based on implementation experience and expert consultation. They are not experimental evidence that following the framework will produce a specific efficiency, cost or AI-performance gain.
