IN Brief:
- Flexible semiconductors could make item-level sensing economical at much larger scale.
- ASICs and FPGAs offer different approaches to processing AI workloads near sensors.
- Thermal design, signal integrity, and connectivity remain major practical constraints.
AI can help generate a circuit, but it cannot make a sealed enclosure shed heat any faster. That tension runs through the October edition of IN Electronics & Design, which follows artificial intelligence from the design workstation into sensors, factories, and everyday products.
Flexible semiconductors offer one way to gather more information from the physical world. Catherine Ramsdale of Pragmatic Semiconductor considers how low-cost chips built into labels and packaging could provide item-level data that conventional tracking often misses. Once captured, however, that information still needs reliable processing and a dependable route into wider systems.
At the edge, the choice of silicon becomes especially consequential. Swindon Silicon Systems examines ASICs designed around specific inference tasks, while Efinix discusses the adaptability of FPGAs as algorithms and workloads change. Grid AI’s assessment of dense sensor boards adds a less forgiving constraint: heat, power transients, and high-speed signals must coexist within very limited space.
The issue also examines the connectivity underpinning industrial digital twins and hears from ByteSnap Design on building automotive Linux systems that can produce camera images within seconds without dispensing with security checks.
Across these applications, performance depends on decisions made well before software reaches the device.
