Sector-Based Learning Ecosystems can Accelerate Transformation for all

To match the speed of change, we need learning infrastructure that doesn’t just deliver training, but helps entire sectors learn, adapt, and transform together.

Last week, Lee mapped out why Europe’s future may hinge on an industrial renaissance, and how AI could help connect and map innovation ecosystems to accelerate development.

“…the most interesting, impactful and possibly also the most economically transformative applications of AI may not be in consumer applications but rather enterprise, industry, government and infrastructure.”

But transformation doesn’t happen through technology alone. It needs human capabilities and ingenuity. To move faster, across manufacturing, construction, energy and beyond, we need to also scale the skills that drive change — and not just company by company, but ecosystem-wide.

The good news is that we already have some models from the last wave of digitisation that we can learn from and build upon. From Fraunhofer Institutes to the UK’s Catapult Network to Bosch’s Agile Academy, early efforts to build modular, distributed upskilling platforms are still developing. The opportunity now is to connect and scale such capability development platforms — and to build on them and design for what comes next.

Most organisations still approach upskilling as something internal, delivered through corporate systems or L&D teams, often lagging a few steps behind strategy. But in a world of accelerating sectoral change, from AI in manufacturing to sustainability in construction, that model is showing its age - because just as products are becoming modular and ecosystems more connected, the same is happening with skills.

When every firm in a sector faces similar capability gaps (e.g. in AI integration, data engineering, digital twins, carbon literacy), it makes less sense for each to build its own solution. Instead, we’re seeing something new: shared learning infrastructure that spans firms, partners, suppliers, and public agencies, and even platforms that scale the skills needed for entire sectors to move forward.

There is also a second shift underway. AI is no longer just a subject of learning, it’s becoming the infrastructure of learning itself. From agentic tutors and digital twins to systems that adapt in real time to skill gaps, AI is turning learning into a live ecosystem capability.

This dual role of AI as both the topic and the enabler, mirrors a deeper shift: from internal transformation to ecosystem innovation; from silos to platforms; from firm-by-firm progress to fast, collective capability upgrades.

The question now is how to design for that.

Why Shared Upskilling Infrastructure Matters

In sectors under pressure to adapt, there is a growing recognition that transformation isn’t just about tools or platforms. It’s about whether the sector as a whole can build and absorb new capabilities fast enough to stay competitive. Shared learning infrastructure is one of the most important enablers of sector-wide capability development. Not because it replaces firm-level training, but because it unlocks something bigger: scale, coordination, and system-level readiness.

When many organisations are grappling with the same capability gaps, in areas like AI engineering, sustainable operations, or agile production, building learning infrastructure collectively has tangible advantages. It reduces duplication, lowers the cost of experimentation, and ensures that knowledge generated through pilots or research doesn't get trapped inside organisational boundaries, or lost when a project loses focus or funding.

This approach also increases the absorptive capacity of the sector - its ability to recognise, adapt to, and deploy new technologies and ways of working. By distributing a common baseline of skills, shared infrastructure shortens the time it takes for innovations to move from early adopters to the wider market. And by embedding learning in live projects or testbeds, it ensures that upskilling is always tied to actual performance and delivery.

Crucially, it also creates positive externalities. Rather than training being a sunk cost borne by each organisation, shared infrastructure creates a skills commons - firms can hire from a more prepared talent pool, SMEs can plug into sector-wide programmes, innovation efforts become easier to scale because they aren’t constantly slowed down by bespoke onboarding or retraining requirements.

But these benefits only materialise when shared learning efforts are treated as infrastructure, not isolated initiatives. That means designing for interoperability, openness, and long-term coordination, and using AI and data to increase responsiveness over time.

At the policy level, this kind of capability infrastructure also supports resilience. A sector that shares its learning models, standards, and digital tools can respond more fluidly to shocks, whether geopolitical, regulatory, or technological.

And with AI starting to augment the learning process itself, through intelligent coaching, simulation-based training, and micro-assessment loops, the impact of shared infrastructure compounds further. It’s not just about spreading knowledge faster. It’s about enabling learning as a system function - an engine that can keep pace with change, rather than react to it after the fact.

What do we have — and what could it become?

Across Europe, the raw materials of a more connected, adaptive learning ecosystem already exist. Institutes, testbeds, and internal academies are doing valuable work, and the EU has funding programmes focused on this challenge.

What’s missing isn’t activity, it’s architecture. The ability to stitch together these efforts into something more programmable, more responsive, and more strategic at the ecosystem level.

The following examples are not perfect models, but they offer clues. And if wrapped in the right networks, communities, tools and services, they could become something greater: infrastructure for capability at scale.

The Fraunhofer Institutes (Germany)

Germany’s Fraunhofer Institutes are often referenced as a model of applied research excellence, but their real strategic value lies in something deeper. They act as infrastructure for capability building at the ecosystem level.

Unlike standalone R&D organisations, Fraunhofer is a distributed network of more than 70 institutes, each focused on a specific domain or sector, but all connected by a common mission: to translate scientific research into practical, industrial value. Their impact stretches far beyond lab walls or academic citations. They’re a core part of how Germany develops, scales, and spreads new capabilities across its industrial base.

In AI, for example, Fraunhofer doesn’t just publish white papers or run pilots. It creates sector-specific pathways for adoption, including training programmes, shared testing environments, simulation labs, and cross-industry demonstrators. Firms, especially SMEs, can access tools, datasets, and expert guidance that would be out of reach if they had to develop them alone. And because the institutes are publicly funded but industry-partnered, they occupy a unique position: close enough to the frontier to influence what’s next, but grounded enough in operational reality to make it usable.

One example: the Fraunhofer IPA (Institute for Manufacturing Engineering and Automation) works directly with manufacturers to explore the use of AI in predictive maintenance, adaptive robotics, and smart logistics. But they also design the learning pathways, training packages, upskilling programmes, interactive workshops, that help firms understand, adopt, and integrate these technologies. This means the learning isn’t an afterthought. It’s embedded in the innovation process from the start.

Fraunhofer’s model is especially valuable in complex, capital-intensive sectors - not because it solves every problem, but because it offers a partial playbook for embedding learning into the innovation process. The challenge now is to move from pockets of good practice to ecosystem-level design: making this kind of capability infrastructure more open, more interoperable, and more connected across sectors and borders.

In a moment where speed, coordination, and absorptive capacity define competitiveness, the Fraunhofer model shows how shared infrastructure can be more than a bridge between research and industry. But to meet the demands of AI-era transformation, we need to go further, building systems that are not only distributed and practical, but programmable and adaptive by design.

Key Metrics on Current Fraunhofer’s Initiatives Impact

• Sales and Productivity Growth: While not focused solely on learning, firms working with Fraunhofer show measurable gains in sales (+9%) and employment (+7%) - evidence of the broader value that shared capability platforms can unlock.

• Return on Public Investment: For every euro of public funding allocated to Fraunhofer, the German economy gains 18 euros in gross domestic product, with 4 euros returning as tax revenue to national, state, and municipal governments.

• SME Satisfaction: Over 80% of small and medium-sized enterprises (SMEs) that have collaborated with Fraunhofer would do so again and recommend the partnership to others.

The Catapult Network: A Platform Approach

The UK’s Catapult Network was established to close the gap between research and industry; but in practice, it’s doing something broader and more strategic: creating a platform for ecosystem-level capability development.

Each of the nine Catapults focuses on a critical domain of the UK economy, from manufacturing and transport to energy systems and digital infrastructure. While their primary mandate is innovation and commercialisation, a quiet but powerful second layer has emerged over time: shared learning infrastructure that supports both technology adoption and organisational change.

The High Value Manufacturing Catapult (HVMC) is a clear example. Headquartered across seven specialist centres, it acts as a distributed engine for industrial transformation, helping UK manufacturers trial new technologies, reconfigure production systems, and build workforce capability in parallel. Programmes span everything from AI in factory automation and predictive maintenance, to advanced materials and circular economy practices. What’s emerging here is not just skills provision, but infrastructure: live, embedded learning environments that operate at sector scale.

HVMC integrates upskilling directly into live testbeds and real-world demonstrators. Firms don’t just learn about digital twins or autonomous inspection, they build and operate them in collaboration with Catapult engineers, then bring the experience and practices back into their own organisations. This fusion of learning and doing helps collapse the usual lag between pilot and adoption, and avoids the common trap where technology readiness outpaces workforce readiness.

The same pattern holds across other Catapults:

• The Connected Places Catapult supports cities and transport authorities to build digital capability through innovation sandboxes and partner training.

• The Offshore Renewable Energy Catapult integrates skills development into collaborative R&D on wind, hydrogen, and marine energy.

• The Digital Catapult provides AI and advanced computing capability development, with SMEs and startups upskilling in the process of co-developing new applications.

What makes the Catapult model powerful is its neutral, pre-competitive positioning. It enables firms to work together on emerging capabilities without giving up commercial advantage, while also inviting public funders, academic partners, and policymakers into the same orbit. This creates a kind of learning commons where resources, methods, and infrastructure can be shared and scaled, rather than constantly reinvented.

And while human-led training still dominates, the foundations for AI-enhanced learning ecosystems are already being laid: simulation environments, capability platforms, and collaborative digital infrastructure are all in place, waiting to be activated with new tooling and design.

The Catapults aren’t yet learning ecosystems in the full sense. But they show how sector-wide infrastructure can seed the conditions for one: modular, partner-driven, and embedded in real work. The next step is to connect and evolve these assets, with AI and ecosystem strategy at the core.

Key Metrics on the Current Catapult Network's Impact

High Value Manufacturing Catapult (HVMC)

• Since 2011, HVMC has collaborated with over 30,000 businesses, including around 3,000 SMEs per year.

• Offers a wide range of workforce development programmes - from apprenticeships to CPD — supporting capability growth alongside tech adoption.

Digital Catapult

• Startups engaged with Digital Catapult raised £555m in funding between 2018–2023, with £172m in 2022/23 alone.

• Facilitated 77 new industry collaborations and delivered 37 digital supply chain projects with 91 SMEs and 267 firms involved.

Bosch's Agile Academy - Scaling from the Inside Out

While national research networks and public-private platforms are critical for sectoral upskilling, some of the most effective capability infrastructure has also emerged from within large firms, especially when those firms treat transformation not just as a technology problem, but as a learning challenge. Bosch is a striking example.

As one of the world’s largest engineering and manufacturing companies, Bosch recognised early that the move toward digitalisation, modular product architectures, and AI-enabled operations couldn’t be delivered by top-down mandates alone. The company needed a shift in mindset, structure, and culture and that required an entirely new approach to how people learn, lead, and collaborate. Having participated in cross-company benchlearning initiatives during their exploration of agile ways of working at the organisation-level, they already understood that learning at scale can be game-changing.

The result was the Bosch Agile Academy, a modular, scalable platform for building the skills, behaviours, and operating model capabilities needed to support continuous transformation. What started as an internal programme to train teams in agile methods has evolved into a broader infrastructure that blends technical, strategic, and cultural learning across disciplines and levels.

The Academy isn’t a static curriculum. It’s a flexible learning system designed to support everything from small, cross-functional teams working on embedded AI projects, to leadership cohorts navigating organisational redesign. It integrates coaching, peer-to-peer learning, and real-world projects - not just in software teams, but across production, logistics, product management, and even HR.

What’s notable is how the Academy has scaled:

• It supports decentralised learning, enabling local teams in different geographies and business units to tailor their development pathways.

• It aligns with Bosch’s broader shift to a platform organisation, where teams work in modular structures that can be recombined around new priorities.

• And it increasingly integrates AI as a delivery layer, using tools to map skills gaps, personalise learning paths, and surface relevant resources based on project context or role.

Bosch has also begun to open elements of this infrastructure externally: running shared formats for partners and suppliers, contributing to cross-industry learning forums, and collaborating with applied research institutions on future-of-work capabilities. In doing so, it acts as a kind of informal anchor institution, modelling what capability-led transformation looks like in practice, and seeding new behaviours and standards across the supply chain.

It’s not a complete ecosystem, but it shows what can happen when capability-building is deeply embedded in how people work, not just an add-on. It becomes a vehicle for organisational agility, and a signal to the sector about what ‘ready for transformation’ actually looks like.

The next challenge, and opportunity, is to connect these kinds of efforts with public infrastructure and emerging AI tooling to create shared, adaptive learning systems that serve broader industrial goals.

Key Metrics on Bosch's Current Upskilling Initiatives

• Training Reach: In 2022, Bosch conducted over 30,000 training seminars worldwide, with more than 520,000 participants. Notably, over 130,000 associates received training in emerging technologies such as electromobility, software engineering, and Industry 4.0.

• Investment in Training: Bosch invested approximately €300 million in workforce training in 2022, underscoring its commitment to continuous professional development.

• Digitalisation of Learning: The company emphasised digital learning platforms, enabling associates to access training independent of time and location, thereby enhancing flexibility and reach.

• External Training Offerings: Beyond internal training, Bosch extended its educational resources externally, offering over 100 training programs on digitalisation and connectivity in manufacturing to other companies and interested parties.

Extending, Not Just Replicating Existing Models

Of course, not every sector shares the same starting point. The structural conditions for transformation - supply chain complexity, regulatory burden, pace of innovation, or capital intensity - differ significantly between, say, advanced manufacturing and construction, or between defence and health. But while the shape of the challenge varies, the design logic behind shared upskilling infrastructure, federated ownership, modularity, public-private alignment, shows up repeatedly.

These patterns don’t offer a one-size-fits-all solution that can be copied. But they do provide a scaffold for building adaptable, resilient learning ecosystems — not as a fixed model, but as a configurable architecture that can flex across contexts. The question is no longer whether we can replicate past success, but how we build on it intentionally, with AI, network effects, and strategic coordination at the centre.

Read on for design patterns for shared upskilling infrastructure and emerging methods that can drive ecosystem learning to the next level.