NVIDIA invested $2 billion in silicon photonics companies Coherent and Lumentum, marking a strategic shift in AI infrastructure design. Photonic interconnects promise higher bandwidth and lower power consumption than electrical connections for chip-to-chip communication in large-scale AI systems.
The photonics push reflects broader semiconductor transformation. Neural processing units now handle specialized AI tasks faster than general-purpose chips, while quantum computing efforts at Intel Labs and Diraq target computational problems beyond classical computing limits. Apple's M4 processor integrates neural engines directly into consumer hardware, demonstrating commercialization of specialized architectures.
Silicon carbide emerged as the standard for high-voltage power systems in electric vehicles. Wolfspeed powers Toyota EV platforms and works with Tier 1 automotive suppliers across "a wide array of EV platforms," positioning the material as foundational to vehicle electrification. The compound semiconductor handles power conversion more efficiently than traditional silicon.
STMicroelectronics launched complete connectivity portfolios for Aliro 1.0, the unified digital access standard. "ST offers the complete secure connectivity portfolio required to support all three Aliro configurations—from NFC only to NFC + Bluetooth Low Energy, up to NFC + Bluetooth LE + UWB for hands-free access," said Luca Verre. The technology enables smartphone-based building and vehicle access.
Grab piloted high-accuracy GPS positioning with OPPO and Swift Navigation in Singapore. Francesco Grilli called the collaboration "an important innovation in bringing high accuracy positioning to mobile devices." The system improves location precision for ride-hailing navigation.
Lattice Semiconductor forecast Q1 revenue between $158 million and $172 million as enterprise customers deploy FPGAs for AI inference workloads. Field-programmable gate arrays offer reconfigurable logic for custom AI acceleration tasks.
The convergence of photonics, neural processors, and compound semiconductors addresses AI infrastructure bottlenecks. Data transfer speeds, power efficiency, and specialized computation capabilities now determine competitive advantage in AI hardware markets. Traditional semiconductor economics face disruption as workload-specific architectures replace general-purpose computing paradigms.