Artificial intelligence (AI) is revolutionising industries, economies, and daily life, with the global AI market projected to grow from $150 billion in 2023 to over $1.5 trillion by 2030. The UK is poised to lead this transformation, aiming to harness AI for societal benefits, economic growth, and global competitiveness through its AI Opportunities Action Plan. However, realising this vision requires a significant leap in hardware development to meet the escalating computational demands of AI systems.
Silicon-based technologies, while advanced, are reaching their limits in terms of energy efficiency and performance. Compound semiconductors, such as silicon carbide (SiC) and gallium nitride (GaN), offer superior properties that can address these challenges. These materials enable faster data processing, greater energy efficiency, and enhanced performance, making them essential for next-generation AI applications.
This report explores the critical role of compound semiconductors in enabling the UK’s AI ambitions. It highlights their applications in power electronics, photonics, and radio frequency (RF) technologies, which are crucial for data centres, edge computing, and the Internet of Things (IoT). By leveraging these advanced materials, the UK can build a sustainable, efficient, and powerful AI infrastructure that drives economic growth and delivers transformative societal benefits.
Introduction
Artificial intelligence (AI) is transforming industries, economies, and the way we live and work. From healthcare and transportation to finance and education, the technology is driving innovation and unlocking new opportunities. The global AI market is projected to grow from $150 billion in 2023 to over $1.5 trillion by 2030 (Statista, 2023).
The UK is at the forefront of this global revolution and has ambitions to become a leader in AI research, development, and deployment. Recognising the potential of AI, the UK has set out an AI Opportunities Action Plan to harness this technology to improve society, drive national growth, and maintain its competitive edge on the global stage.
However, to fully realise the benefits of AI, a step-change is needed to develop the advanced hardware to support the huge computational demands of machine-learning algorithms and their supporting systems.
Silicon semiconductor technology has made remarkable strides in recent years, with companies like Nvidia developing cutting-edge solutions and leading the way. However, as AI applications grow in complexity and scale, silicon-based hardware are reaching their limitations.
Training a single large AI model consumes as much energy as 100 homes use in a year (The Verge, 2024), and data centres already account for approximately 1% of global electricity demand (IEA, 2022). AI hardware and infrastructure must therefore become faster and more energy-efficient.
This is where compound semiconductors can play a unique role. With certain properties that are superior to traditional silicon, compound semiconductors present a transformative opportunity to address the challenges of AI hardware.
Power electronics based on compound semiconductors can significantly enhance the energy efficiency of the data centres that power our AI. Data centres are projected to consume up to 8% of global electricity by 2030 if current trends continue (IEA, 2022).
Photonic devices built using compound semiconductors will dramatically increase the speed at which computers process and transmit data, significantly reducing latency in AI applications.
Meanwhile, radio frequency (RF) technologies leveraging compound semiconductors will improve the speed and efficiency of data transfer in AI applications, edge computing, and the Internet of Things (IoT), which is expected to connect over 32 billion devices worldwide by 2030 (Statista, 2024).
Compound Semiconductor Applications (CSA) Catapult is the UK’s authority on compound semiconductor applications and commercialisation.
As a trusted neutral convener of academia and industry, CSA Catapult is uniquely positioned to drive the development of the hardware needed to power the UK’s AI action plan.
This report explores the critical role of compound semiconductors in enabling the UK’s AI ambitions, highlighting the opportunities they present and recommendations for developing and integrating these technologies into the AI ecosystem.
By leveraging the unique capabilities of compound semiconductors, the UK can build a future where AI not only drives economic growth but also delivers transformative benefits to society.
AI hardware landscape
AI workloads are expanding at an unprecedented rate. As models become larger and more sophisticated, the computational power required to train and deploy them is increasing exponentially. Since 2012, the computational resources needed to train state-of-the-art AI models have doubled every 3.4 months (OpenAi, 2018).
The trend shows no signs of slowing, with AI applications being integrated into everything from autonomous vehicles and smart cities to healthcare diagnostics and industrial automation.
As a result, AI-specific semiconductor chips have experienced soaring demand. AI semiconductor revenue is forecast to be $196.5 billion by 2028, growing by a fiveyear CAGR of 29.6% (Gartner, 2024).
The rising compute demand of AI will create additional demand for many other components and chip types, including power semiconductors and optical interconnects. These will be made using compound semiconductors and will be primarily located in data centres and at the edge.
Data centres are the backbone of AI infrastructure, housing the vast computational resources required to process and store the enormous amounts of data generated by AI systems.
As AI adoption accelerates, the number of data centres is expected to increase exponentially.
Globally, there are over 10,500 data centres in operation, including more than 1,000 hyperscale facilities. Over half of the world’s data centres are in the US, who have built an additional 5,000 in just seven years at compound annual growth rate (CAGR) of 44% (CSA Catapult, 2025).
The UK has 514 data centres – the third largest amount in the world (CSA Catapult, 2025). By 2030, global data centre energy consumption could account for up to 8% of the world’s electricity demand, up from 1% today (IEA, 2022).
The rise in energy use is being driven by the need to power and cool the servers that run AI workloads, which are among the most energy-intensive tasks in computing.
A recent report by the Royal Academy of Engineering urged the UK government to promote, prioritise and invest in sustainable AI and recommended the government make it a requirement for tech companies to report on how much water and energy their data centres are using.
As AI workloads continue to grow, the energy efficiency of hardware will become a key differentiator in the market.
Current silicon-based technologies are struggling to meet these dual demands. While advancements in silicon processors have delivered impressive performance gains, particularly on the compute and memory side of AI, they are increasingly constrained by power consumption and heat dissipation issues.
Compound semiconductors offer a unique opportunity to address these challenges, delivering the performance and efficiency needed to power the next generation of AI applications. By investing in advanced hardware solutions, the UK can position itself as a global leader in AI innovation while ensuring that its AI infrastructure is sustainable and future-proof.
