Innovation Districts on the Urban Edge: Foundational concepts, case lessons, and a policy architecture

John H. Howard[1], 21 July 2026

This working paper is one of a series prepared by the Acton Institute for Policy Research and Innovation to inform the UTS research project Turning AI into Productivity: The Role of Innovation Ecosystems, supported by the Google Foundation. The project's final report will be launched in mid-September 2026.

A recurring question in innovation policy is where innovation ecosystem activity should concentrate and how it should be organised. Internationally, a discernible pattern has emerged over the past half century.

Many of the districts that turn research into production grow on the edge of the city rather than in its dense centre, on land with room to expand. These edge-of-city districts organise themselves around one of three anchors: public infrastructure, a university campus, or an advanced-industry base.

The 2026 UTS research project Turning AI into Productivity: the Role of Innovation Ecosystems examined a set of them across Europe and the Nordics, and the same structural features recur from Eindhoven to Espoo.

Australia now has candidates of its own. Tonsley in Adelaide converted a closed car plant into a working district, and potentially, Bradfield, the commercial centre of the Western Sydney Aerotropolis, is being built beside a new international airport.

This Insight brings the foundational concepts, the case lessons, and the policy architecture into a single account and treats these Australian precincts as members of the same family.

Foundational Concepts

Five concepts from the Handbook of Innovation Ecosystems frame the analysis. They are tools for thinking, not rules to apply, and each carries a direct implication for how a district might be designed, funded, and governed.

1. Ecosystems, not clusters

The Handbook draws a line between an innovation ecosystem and an industrial cluster. A cluster gathers similar firms in one place. An ecosystem is a web of complementary players, knowledge institutions, firms, investors, intermediaries, and government, whose interactions produce capabilities no member holds alone. A district planned as a cluster tends to leave that supporting web unfunded.

2. The Complementarity Principle

The Complementarity Principle holds that an asset becomes productive through the complements around it. Equipment, skills, data, finance, and management practice reinforce one another. A district with a strong set of complements can outperform one that owns a larger single asset but lacks the supporting pieces. The design task is to build the whole set rather than chase a single part of it.

The reasoning extends to anchors. A technology or facility holds no productive identity of its own; it acquires one from the assets around it. An airport by itself does little. Add logistics firms, skilled workers, research links, and steady demand, and it becomes a working hub. The anchor does no real work until its complements arrive.

3. Generative and connective anchors

This reasoning helps separate two kinds of anchor. A generative anchor, whether a research facility, a major employer, or an applied institute, gives a district its reason to exist and creates the activity people travel for. A connective anchor, whether an airport, a metro station, or a motorway interchange, provides reach, widening the catchment so the other anchors can do their work.

Figure 1. Generative and connective anchors

Edge-of-city districts lean on the connective kind more than central ones do. A dense centre supplies cheap proximity, the repeated encounter that builds trust, while a fringe district has to manufacture it. Efficient public transport is the means, and it reaches knowledge workers who may choose to live elsewhere, near a partner's work, family, or schooling, rather than relocate to the precinct.

Knowledge workers are also drawn to cultural amenity, the galleries, venues, food, and street life that supply creative stimulation and chance encounter. A new precinct on the fringe cannot offer this at once. Good transport lets workers reach the cultural life of the wider city while the district's own placemaking builds it over time.

4. The Management Gap

Many districts hold strong research alongside strong industry but cannot connect the two. The gap between discovery and production is the Management Gap, and promising ideas stall in it for want of translation capacity. Closing it takes intermediary institutions, applied facilities, skilled translators, and routines that carry work from the laboratory towards the factory floor.

Figure 2. The Management Gap and the role of translation capacity

5. Four-Domain Convergence

Drawing on the logic in the Handbook of Innovation Ecosystems, a district works when four domains converge. Placemaking gives the place its quality and identity. Economics covers jobs, value, and investment. Business is about firm formation, growth, and supply chains. Governance supplies the rules, the vehicles, and the leadership that hold the rest together. Each domain leans on the others, so a shortfall in one drags down the whole.

Figure 3. Four-Domain Convergence

What Current Edge-of-City Districts Tell Us

The districts studied differ in history and sector, but share structural features that recur wherever the pattern appears. The primary European and Nordic cases below carry most of the argument, with Tonsley in Adelaide as an Australian comparison.

Stockholm, Kista, and Hagastaden

Kista Science City, on the northern edge of Stockholm, grew up around Ericsson and the wider telecommunications sector, with KTH Royal Institute of Technology supplying research and graduates through its campus in the district. For three decades it stood as the model of an anchor-led district, marketed as Sweden's Silicon Valley.

That model is now being tested. In May 2026 Ericsson announced the relocation of its Stockholm operations, including its global headquarters, from Kista to Hagastaden on the northern edge of the inner city. The move begins in 2028 and runs in phases over several years, on a city campus of about 95,000 square metres under the largest office lease recorded in Sweden.

The company framed the decision around access to partners, decision makers, and future talent. Hagastaden sits beside the Karolinska Institute and its university hospital, a short distance from KTH's main campus, within the band of knowledge institutions along the city's northern inner edge. Ericsson is leaving a district where it was the generative anchor for one where the complements are already assembled.

An offshore move has also been contemplated in late 2024. The company's chief executive said that relocating the headquarters to the United States “could well happen”, citing a European telecommunications market that was falling behind. The remark was widely read as pressure on European regulators rather than a settled intention, and the company did not follow through.

Several factors appear to have kept the company in Stockholm: its ownership is anchored in Swedish institutional investors, its engineering pipeline runs through KTH, and North American sales weakened in early 2026. The succession announced in mid 2026 passed to a Swedish company veteran educated at KTH. The complements that matter most to the firm are assembled in Stockholm and are not readily portable.

TU Dortmund

TU Dortmund demonstrates the conditions under which an outer-edge concentration matures into an ecosystem: co-located and interdependent institutions rather than a lone university, high-frequency transit as a substitute for urban adjacency, on-site translation capacity, purpose-built amenity, and patient metropolitan governance that treats the edge site as part of a distributed city system rather than as a stand-alone project. Absent those conditions, an outer-edge campus remains an island.

Eindhoven and Brainport

Eindhoven rebuilt its economy after Philips and DAF contracted. The High Tech Campus, laid out on a former Philips research site, became an open innovation campus, while a triple-helix development company steers industry, knowledge institutions, and government towards a shared agenda. ASML now anchors a dense semiconductor supply chain. Patient coordination, sustained over decades, made the difference.

Munich, Garching, and Heilbronn

The Technical University of Munich runs campuses on the city's edge at Garching and, more recently, at Heilbronn, the latter backed by philanthropic capital and sitting beside an artificial-intelligence innovation park. The model pairs a research university with entrepreneurship support and long-term funding, and produces firms and talent at scale. Philanthropic and institutional money can carry a district where public budgets alone might not.

Otaniemi and Espoo

Otaniemi sits in Espoo, on the western edge of the Helsinki region and a short metro ride from the centre. It brings together Aalto University and VTT, the largest applied research institute in the Nordics, with a startup community that gained momentum as Nokia shed talent during its decline. The case pairs a translation institute with the talent and ideas a large employer sheds when it contracts.

Tonsley, an Australian comparison

Tonsley, about ten kilometres south of Adelaide, was rebuilt from the closed Mitsubishi car plant into an innovation district led by the state agency Renewal SA. Flinders University and TAFE South Australia sit beneath the retained assembly building roof, alongside firms in advanced manufacturing, clean energy, and medical devices. As the nearest Australian precedent, it shows an industrial site becoming an ecosystem, given patience and sustained public commitment.

The common pattern

The same features recur across the cases: an edge-of-city location with room to grow, a research anchor, one or more industry anchors, a dedicated governance or development vehicle, translation institutions, and a horizon measured in decades. None of these places succeeded quickly, and none succeeded on property development alone.

Most also rest on a rail or metro connection to the wider city, a feature that has grown more prominent over time. The early science parks, such as Research Triangle, Stanford/Silicon Valley, and Sophia Antipolis, were built for the car on generous land, while later districts, such as Otaniemi, Kista, Pangyo, and Adlershof, rely on public transport as talent preferences and congestion have shifted.

The Wider Family

A broader survey extends the pattern across Europe, Asia, and North America, from the earliest science parks to recent planned tech valleys. Each grows on a city's edge, organised around a research anchor, a campus, or an advanced-industry base, and each is shaped by a development body with a long mandate.

The table below gathers a broader set, each case adding a distinct lesson to the common pattern rather than repeating it.

District (city/edge)	Anchors	Development or governance vehicle	What it adds to the pattern
Jurong Innovation District, Singapore (western edge)	Nanyang Technological University; A*STAR institutes; Siemens, Hyundai	JTC, the industrial-estate statutory board (phased build-out)	A state developer assembling research, production, and talent on one campus, with living-lab piloting.
Hsinchu Science Park, Taiwan (1980)	Tsing Hua and Yang Ming Chiao Tung universities; ITRI, which seeded TSMC, UMC, and MediaTek	A dedicated park administration for land, incentives, and approvals	A translation institution and patient public backing building a whole sector from a standing start.
Leuven, east of Brussels	KU Leuven; imec, the nanoelectronics institute	Flemish government (imec founded 1984)	A neutral commons where rival firms share advanced facilities, producing hundreds of spin-offs.
Harwell, Oxfordshire (about 25 km south of Oxford)	Diamond Light Source and the Rutherford Appleton Laboratory; a dense space cluster	A former RAF and atomic-research site holding national facilities	Shared national equipment and a defence and space mix relevant to advanced-industry precincts.
Ideon, Lund, Sweden (1983)	Lund University; MAX IV and the European Spallation Source nearby	Sweden's first science park	National research facilities anchoring a district over the long term.
Adlershof, Berlin (south-east)	Humboldt University science faculties; research institutes; more than a thousand firms	State-owned WISTA, after reunification	Housing planned from the start, with the S-Bahn tying the site to the labour market as residents and firms arrived in sequence.
Cambridge, on the edges of a historic core	Cambridge Science Park, the Biomedical Campus, and the Institute for Manufacturing	University of Cambridge	Research capacity spread to the edge, with a translation function set apart from the colleges.
Daedeok Innopolis, Daejeon, Korea (1970s)	KAIST; government laboratories; more than a thousand ventures	A national science-city plan	A research university and public laboratories anchoring a national R&D cluster.
Pangyo Techno Valley, Seongnam, Korea (2011)	Naver, Kakao, and games, internet, and biotech firms	Gyeonggi Province (land, infrastructure, tax incentives)	A provincial government assembling a high-value cluster on the fringe within a decade.

Read together, the cases point to a few recurring lessons. State developers can assemble research, production, and talent on a single campus. Translation institutions such as ITRI, imec, and the Institute for Manufacturing act as the connecting tissue. Housing, transport, and research work best when sequenced, and most of these districts took decades to mature.

A Policy Architecture for District Success

From these cases, a policy architecture for a successful innovation district emerges. It is a set of interacting components that have to be present, resourced, and put in the right order, not a checklist to tick off. The components below are meant to frame deliberation, not to prescribe a course for any one precinct.

• An anchor with pull. A research facility, major employer, or piece of infrastructure that gives the district a reason to exist and draws talent and investment towards it.

• A translation institution. A body whose job is to move ideas into production through applied research, testbeds, demonstration facilities, and shared equipment. It is the piece most often missing where districts underperform.

• A governance vehicle with a long mandate. A coordinating entity, shielded from short political cycles, able to assemble land, capital, and partners over decades.

• Patient capital. Public and private funding that accepts long payback periods, philanthropic and institutional sources among them. Heilbronn shows one route, and blended public-private arrangements another.

• A talent pipeline. Education and training tied to the district's industries, with pathways from school through apprenticeship to advanced research, designed alongside employers.

• Early demand and procurement. Customers, government included, that give firms a reason to locate and to grow. Defence and advanced-manufacturing procurement may both apply.

• Connectivity and accessibility. Efficient public transport that ties the district to the regional labour market, to research institutions, and to where its workers live. Many will settle near a partner's work, family, or schooling rather than relocate, so the district has to reach them.

Figure 4. A policy architecture for district success

These components reinforce one another, and the order in which they arrive matters. Early investment in governance, anchors, and translation capacity tends to repay itself well before population and amenity catch up. A district built for property returns alone may fill its floor space while never building the capabilities that keep it alive.

A Cautionary Note: the Adelaide Multifunction Polis

The Multifunction Polis offers a counterpoint. Proposed in 1987 as a futuristic Australia-Japan city, and given a site at Gillman on Adelaide's north-western fringe in 1990, it never moved beyond planning. A racially charged campaign against a supposed Japanese enclave, the collapse of the Japanese asset bubble, and a scope that kept shifting drained its support.

Federal funding ended in 1996, and the project was abandoned with little built and about $150 million spent. A working precinct did emerge later and a short distance away, as Technology Park Adelaide and the Mawson Lakes development. The lesson is that a credible development vehicle, patient sequencing, and social licence matter more than a visionary masterplan.

Where the Analysis Leads

The record points to a few durable conclusions for anyone building or governing an edge-of-city district. The generative and connective anchors both need their complements assembled deliberately, since neither becomes productive on its own. Kista's experience adds a further caution: an anchor can leave, and a district needs more than one reason to exist. Translation capacity is the component most often underfunded, and its absence is felt where research and industry sit side by side without connecting.

Governance carries the weight over time. A vehicle with a long mandate, patient capital, and social licence outlasts the political and market cycles that a masterplan cannot survive. Connectivity has moved from a convenience to a condition, since talent increasingly reaches a fringe district by rail rather than relocating to it.

Australian precincts sit inside this international family rather than apart from it. Tonsley shows an industrial site becoming an ecosystem given patience, and Bradfield holds several components in early form: an infrastructure anchor in its airport, an applied facility, a governance vehicle, and university links across the region. The metro gives it a connective anchor early, an advantage over districts that retrofitted access. The open questions everywhere concern depth, sequencing, and the parts not yet in place, and each is far easier to settle early than to revisit late.

References

Blandy, R. (2017). SA’s big bang theory: Lessons from inside the MFP. InDaily.

Bradfield Development Authority. (n.d.). Bradfield City Centre. NSW Government. Retrieved June 21, 2026.

Brainport Eindhoven. (n.d.). What is Brainport Eindhoven. Retrieved June 21, 2026.

City of Espoo. (n.d.). Otaniemi: A unique innovation hub. Retrieved June 29, 2026.

Diamond Light Source. (n.d.). About Diamond. Harwell Science and Innovation Campus. Retrieved June 29, 2026.

DTU Science Park. (n.d.). About DTU Science Park. Retrieved July 4, 2026.

Ericsson. (2026, May 25). Ericsson to relocate Stockholm operations to Hagastaden [Press release].

Hovedstadens Letbane. (n.d.). The light rail along Ring 3. Retrieved July 4, 2026.

Howard, J. H. (2025). Handbook of innovation ecosystems. Acton Institute Publishing.

Howard, J. H. (2026). Making sense of AI in 2026. Acton Institute Publishing.

Hsinchu Science Park. (n.d.). About HSP. Retrieved June 29, 2026.

Ideon Science Park. (n.d.). Who we are. Retrieved June 29, 2026.

Infrastructure NSW. (2025). Aerotropolis Sector Plan. NSW Government.

Institute for Manufacturing. (n.d.). Institute for Manufacturing (IfM). University of Cambridge. Retrieved June 29, 2026.

JTC Corporation. (n.d.). Jurong Innovation District. Retrieved June 29, 2026.

Korean Culture and Information Service. (2023). Daedeok Innopolis, a city building the future of science. Retrieved June 29, 2026.

KU Leuven. (n.d.). Leuven, knowledge pearl. Retrieved June 29, 2026.

Light Reading. (2026, June 17). Ericsson appoints Per Narvinger as CEO to replace Ekholm.

Mobile World Live. (2024, November 21). Ericsson boss opens door to future US relocation.

NSW Government. (2024). Bradfield City Centre Master Plan. Bradfield Development Authority.

Pangyo Techno Valley. (n.d.). Pangyo Techno Valley. Retrieved June 29, 2026.

Renewal SA. (n.d.). Tonsley Innovation District. Government of South Australia. Retrieved June 29, 2026.

Research Triangle Foundation of North Carolina. (n.d.). Research Triangle Park. Retrieved June 29, 2026.

Sophia Antipolis. (n.d.). The technopole. Retrieved June 29, 2026.

Stanford Research Park. (n.d.). About. Retrieved June 29, 2026.

The Next Web. (2026, May 25). Ericsson is leaving Kista for central Stockholm, in the largest office lease in Swedish history.

Universite Paris-Saclay. (2025). The Universite Paris-Saclay university innovation cluster (PUI). Retrieved June 29, 2026.

WISTA Management GmbH. (n.d.). Science and technology: Overview. Technology Park Berlin Adlershof. Retrieved June 29, 2026.

[1] Dr John H. Howard is Executive Director of the Acton Institute for Policy Research and Innovation, Sydney, and Research Director of the UTS research project Turning AI into Productivity: The Role of Innovation Ecosystems, supported by the Google Foundation. The project is led by Emeritus Professor Roy Green AM, Special Innovation Adviser at the University of Technology Sydney, who reviewed drafts of this paper and contributed advice on its argument and policy framing.

The paper draws on research undertaken for the project, including visits to 14 innovation ecosystems across eight European countries in May and June 2026. Responsibility for the views expressed, and for any errors or omissions, rests with the author.