The semiconductor industry is at the edge of a new discontinuity

In 2010, Apple introduced its first custom-designed processor – the A4 chip. Today it has more than 6,000 engineers working on chip design. Tesla, looking for semiconductors capable of supporting the huge processing demands of its electric vehicles and autonomous driving systems, has taken the same approach. In 2019, it revealed its own AI chip for its Autopilot hardware. AWS has designed its own Graviton processors, because it wanted to offer better price-performance ratios for cloud customers and what was available on the market could simply not deliver on these requirements.

The shift from a complex semiconductor supply chain to a vertically integrated model of innovation – whereby companies own their own processor, integrated circuit design and development process – represents a seismic shift in the industry. Conjuring images of the 1960s, when technology behemoths like IBM owned the entire technology stack, companies across many different sectors are pulling this all in-house to gain a competitive advantage, by creating chips that are precisely tailored to their needs – improving hardware/software partitioning, performance, power efficiency, supply security and differentiation. But, unfortunately for most OEMs, that is simply not an option they can afford.

R&D costs are surging

The cost associated with advanced chip design – encompassing hardware, software, prototyping, qualification,  and application – has witnessed an astronomical surge. A decade ago, developing a 28nm chip cost roughly $50 million. Today, the price tag for cutting-edge chip design is more than 20 times higher. This surge extends to electronic design automation (EDA) tool costs, further complicating the landscape. So, whilst the very biggest companies in the world can spin up their own teams of engineers to do R&D, the majority cannot carry that cost. 

As a result, whilst advanced chip manufacturing largely remains locked to the foundries of TSMC, Samsung and Intel, it is in the research, development, and design process where we are seeing the emergence of new business models to address the need for OEMs to become more vertically integrated in a cost-effective way. That creates fresh opportunities for exciting growth. For example, with chiplets emerging as a modular solution, there is a growing appetite to separate the design, manufacture, and testing of processor components, offering unparalleled flexibility and cost-effectiveness at the design stage.

New collaborative models

One of the most interesting new approaches is the collaboration between OEMs and strategic technology partners for semiconductor research and design. Under this model, the technology partner acts as the OEM’s external semiconductor department or co-owned department, or it engages in co-design activities and the development of co-owned solutions. This approach removes the cost barrier for building an in-house research, development and design function, opening up the option for more companies that can therefore enjoy the benefits of vertical integration outlined above.

It suggests that the future of the semiconductor industry may lie in a service-oriented business model and collaborative design approaches. OEMs will need to engage in co-design and co-ownership partnerships to navigate the evolving landscape and build a strong ecosystem effectively.

That is the change that we foresee – the creation of an industry environment that drives the development of ecosystems between foundries, IP providers, EDA companies, software developers, and OEMs, shaping the future of silicon service engineering in Europe and beyond.

Does Europe need a new approach?

Europe is at the heart of these challenges. Why? Because there is a burning need for collaboration across the continent. There are no more than a handful of European companies that can afford these advanced nodes. Europe must aggregate its strengths to take on that challenge. The good news is that Europe fundamentally has the required technology, talent, resources, projects, and expertise. Promising companies like Kalray, SiPearl, VSora, GreenYellow, Menta, Scalinx, and GrAI Matter Labs and many others offer real hope, but only if we can develop a collaborative ecosystem in Europe that encourages OEMs to partner with companies like this as part of their own move to a more vertical strategy.

This a major challenge for the European Chips Act (ECA) if it is to help bolster Europe’s semiconductor industry. It’s about enabling a new model to match the vertical integration championed by the largest companies in the world, at a fraction of the cost. That is going to require a significant amount of collaborative work to put in place the incentives and frameworks that can support such a seismic shift in a short space of time. With this properly established, European companies will have a fighting chance to successfully gain ground in the semiconductor race.