Cardiff and Newport offer a commercially meaningful compound-semiconductor corridor because research, wafer activity, manufacturing equipment, prototyping and applications expertise exist within the same regional system. That does not make the chain complete or low-risk. A company must identify the exact process capability it can access, the customer application it serves and the ownership, supplier and buyer dependencies that could interrupt growth.

The right test is movement through the value chain. University intellectual property must become a manufacturable device, the device must enter a customer's system, and the company must finance qualification and repeat supply. Cluster membership is useful only where it makes one of those transitions faster or more reliable.

Start with the semiconductor value chain

Semiconductors are not one market. The chain includes materials and wafers, device design, fabrication, packaging, testing, manufacturing equipment, software and downstream systems. Compound semiconductors use materials other than conventional silicon for properties suited to applications such as radio frequency, photonics, sensing and power electronics.

The UK National Semiconductor Strategy identifies compound semiconductors, design and intellectual property, and research as UK strengths. Its cluster map places South Wales alongside other specialist regions. The strategy provides national context, but company decisions need current facility and market evidence.

A regional map should assign each organisation a verified role rather than place all logos under one heading. It should distinguish research, design, epitaxy or wafer work, fabrication, tools, packaging, testing and application development. It should also mark ownership and headquarters because local operations can be affected by decisions taken elsewhere.

The comparison with Bristol semiconductor design and London capital is useful precisely because Bristol occupies a different part of the chain.

University IP needs a route into process development

Cardiff and other regional universities contribute research and trained people. The commercial question is whether a company can secure the relevant intellectual property, reproduce the result and transfer it into an industrial process.

The UKRI Gateway to Research can identify publicly funded projects and partnerships. The Intellectual Property Office patent search can help trace published inventions and ownership. Neither source shows the full body of confidential know-how needed to make a device.

A founder should document background rights, process recipes, data, software, mask or design assets, test methods and rights to improvements. It should identify which elements are controlled by a university, company, sponsor or facility operator. This work determines whether the business can license technology broadly, work in only one field or depend on a partner for continuing access.

The Sheffield article on what happens after advanced manufacturing research leaves the lab provides a general commercialisation test. Semiconductors add costly process transfer, equipment and qualification dependencies.

Regional facilities can reduce the distance between design and manufacture

The CSconnected programme page describes an effort to integrate university research with regional advanced-semiconductor supply chains. It identifies collaborative research, skills, cluster coordination and facilities as programme components. Several published benefit figures are projections and must not be reported as achieved outcomes without an evaluation.

The Compound Semiconductor Applications Catapult provides facilities and expertise intended to support development and commercialisation. A company should confirm which services are available, the technical specifications, queue, price, confidentiality and ownership terms.

Open access is not a binary label. A facility may accept external work but still be unsuitable for a process, volume, material or customer requirement. The founder should obtain a written process flow showing which steps can be completed locally and which must move to another UK or overseas supplier.

The Cardiff Capital Region Investment Zone entered its delivery phase after an agreement reported in 2026. Its private-investment, job and floorspace figures are projections over a programme period. They should be tracked as approved, contracted, built and operating rather than collapsed into one announcement.

Applications and customers determine which capability matters

A compound-semiconductor device earns value inside a downstream system. The cluster's public materials reference communications, electric vehicles, power and medical applications. A company must narrow that list to a buyer, specification and qualification route.

The Semiconductor Sector Study 2026 provides a current national view of regional capabilities, including South Wales. It should be used to frame questions, not substitute for customer evidence.

Qualification can take a long time because the device may affect safety, reliability or performance over the life of another product. The supplier needs clear test standards, sample volumes, traceability, failure analysis and change control. It also needs to know whether the first customer has authority to nominate the component for production.

A research partner, demonstrator or evaluation agreement is not necessarily a customer. The strongest evidence is a qualified part, repeat order or publicly verifiable supply relationship. Private contracts may remain confidential, in which case the article should state what cannot be confirmed.

London adds capital and specialist services, not semiconductor physics

London can provide venture and growth capital, corporate finance, insurance, legal advice, public-market expertise and access to some customers. It cannot replace fabrication, process engineers or application testing. A South Wales company may therefore maintain investor and commercial relationships in London while keeping technical control close to Cardiff and Newport.

The amount of London presence should follow the frequency of work. Regular fundraising, partnerships or senior recruitment may support a small team. Occasional transactions may require only travel and advisers. Moving the technical organisation can weaken the very cluster advantage that made the company invest in South Wales.

Capital type matters. Semiconductor companies may need grants for research, equity for product development, equipment finance and working capital for long qualification cycles. Each instrument should be connected to a technical or commercial milestone.

Public filings from Companies House can show accounts, charges and legal ownership. They cannot, by themselves, show production volume, local employment or customer concentration.

Concentration risk is part of the cluster story

Clusters create efficiency through proximity, but concentration can also create shared risk. A company may depend on one foundry, equipment provider, specialist workforce, customer sector or parent-company decision. Supply interruption or ownership change can affect several firms at once.

The regional map should therefore include alternative suppliers, process-transfer options and critical equipment. It should record which capabilities are unique, which have UK alternatives and which depend on overseas services. This is commercially more useful than a promotional count of organisations.

Customer concentration also matters. A large contract can validate a technology while making the supplier vulnerable to one programme. The board should model delayed qualification, lower volume and the loss of its largest buyer.

National security and export rules can affect ownership, transactions and customer markets. These issues require specialist legal advice and current checks. They should not be reduced to general claims about strategic technology.

A site decision should follow the missing link

A research spinout should locate close to the facility and expertise needed for process transfer, then confirm a path to production. An established semiconductor company should compare local equipment, people, customer support and incentives with the cost of integrating a new site. An applications business may benefit from Cardiff and Newport only if the cluster solves a specific device, testing or supply problem.

The decision sheet should list process steps, facility owner, access status, alternative supplier, lead time, cost and intellectual-property terms. It should also name target applications and qualification gates. This turns the corridor from a narrative into an operating plan.

Limitations

Semiconductor ownership, production and customer relationships can change quickly. Public cluster pages mix existing capability, programme activity and projected outcomes. Company filings lag and may not separate South Wales operations. Private utilisation, yield, qualification and customer data are rarely available.

Before publication, every named company and facility requires a current operating and ownership check. The proposed supply-chain map needs technical review, customer evidence and a clear evidence date. Investment Zone forecasts must remain separate from completed buildings, operational capacity, jobs and private investment.

Reporting by Shoreditch Talk