Custom software and silicon set to define next-gen chips

Confidence in the semiconductor industry’s ability to meet demand is being hindered by geopolitical tensions, international trade restrictions and the push for sovereignty, according to a report from Capgemini. 

Authors of The semiconductor industry in the AI era report warn that as nations compete for control over vital technologies and resources, geopolitical tensions continue to impact the global semiconductor supply chain.

For instance, the flow of components, materials and completed semiconductor products has been hindered by international trade disputes, export restrictions and tariffs. As an example, Capgemini estimated that 14 consumer supply networks that rely on TSMC could be seriously disrupted by any military escalation involving China and Taiwan. Deteriorating US-China ties have also given rise to setbacks in the form of prohibitions on certain products and more stringent controls.

The report, based on a survey of 250 semiconductor firms and 800 “downstream” organisations that use semiconductors in their own products, found that 58% of semiconductor organisations expect higher demand for neural processing units to accompany growth in generative AI (GenAI) adoption. The study also shows that 57% of the chip manufacturers surveyed anticipate an increased need for high-performance chips, and 56% expect to see greater demand for memory-intensive chips, which, according to Capgemini, signals a shift towards advanced processing.

The report shows there is increased demand for custom chips and custom software optimised to run on them.

Jiani Zhang, executive vice-president and chief software officer at Capgemini Engineering, said advanced platforms and software are critical differentiators in the semiconductor industry, driving efficiency and scalability in design, manufacturing and deployment.

“With the growing complexity of AI, IoT [internet of things] and edge computing applications, the ability to integrate domain-specific software with hardware accelerators will define leadership,” she said. “To stay competitive, semiconductor players must embrace co-optimisation across the stack, from chip architecture to application interfaces, ensuring they can meet the escalating demands of data-intensive, low-latency markets.” 

Capgemini said that while the demand for AI chips, custom silicon chips and memory-intensive chips is expected to increase over the next 12 months, the semiconductor industry needs to capitalise on emerging opportunities. These include design and cutting-edge, sustainable fabrication methods, as well as investment in domestic sourcing and nearshoring to enhance stability.

“GenAI is driving accelerated demand for chips, and semiconductor companies face increasing demands from customers who want more personalised and software-centric experiences,” said Brett Bonthron, global high-tech industry leader at Capgemini.

“The industry should see this as an opportunity to ramp-up production and adopt a ‘chip-to-industry’ approach that supports a full-stack, ‘software-first’ set of capabilities,” he added.

“Investment in cutting-edge fabrication methods and design processes powered by AI and GenAI will be key to meet the specialised needs of emerging applications. Equally, it’s crucial that the industry further enhances sustainable manufacturing processes and uses advanced security to safeguard intellectual property.”