Soitec (SOI) AI 2026 presentation summary

AI fundamentals and market evolution

• AI complexity is rapidly increasing, driving exponential growth in computing power demand, especially for training large models like GPT-5 and Grok 4.

• AI is evolving from data understanding and content generation to agentic and physical AI, with future systems expected to interact autonomously in the real world.

• The shift from cloud-only AI to edge AI is underway, emphasizing lower latency, higher energy efficiency, and privacy by design.

• Edge AI silicon revenue is forecasted to grow at a ~15% CAGR through 2030, driven by IoT, endpoint, and infrastructure applications.

• AI models are increasingly deployed both in hyperscale data centers and on-device edge chips, each with distinct hardware and substrate requirements.

Engineered substrates for cloud AI

• Data center capacity and capital spending for AI are projected to grow at over 20% CAGR through 2030, straining bandwidth, power, and energy efficiency.

• Optical interconnects are essential for overcoming bandwidth bottlenecks, with a shift from pluggable optics to co-packaged optics (CPO) and photonics interposers.

• Silicon photonics, built on SOI substrates, is the dominant platform for high-speed optical interconnects, offering scalability, integration, and cost advantages.

• LNOI (Lithium Niobate on Insulator) substrates target energy-efficient, high-speed data transfer for advanced data center and quantum applications.

• FD-SOI substrates enable scalable quantum computing by integrating quantum and classical transistors on a mature, CMOS-compatible platform.

Engineered substrates for edge AI

• Edge AI devices require high-density memory, ultra-low power, robustness, and optimized AI models, with FD-SOI substrates meeting these needs.

• FD-SOI is widely adopted for embedded and low-power edge computing, supporting applications from wearables to smart home devices and medical IoT.

• Wearables and smart glasses benefit from FD-SOI's energy efficiency, always-on performance, and compact integration, enabling longer battery life and advanced features.

• The product portfolio is expanding to include substrates with various active layers (e.g., Si, SiC, GaN, GaAs, InP) to address future AI, RF, and optoelectronic applications.

• AI-nabler substrates are designed for better performance, higher yield, and greener technologies across edge and cloud AI markets.