FotoNation Ltd., a European perception‑recognition specialist, announced a strategic partnership with SEMIFIVE Inc., the South‑Korean AI‑chip foundry known for custom ASIC solutions. Under the agreement, SEMIFIVE will lead the full development cycle—from architecture to tape‑out—of FotoNation’s TriSilica chip family. The deal marks SEMIFIVE’s first European contract, expanding its footprint beyond the United States, China, Japan, and India.
What is TriSilica?
TriSilica is an ultra‑low‑power, compact‑footprint AI silicon platform that ingests multimodal sensor data—audio, mmWave, spectral, infrared, and RGB—in a single SoC. The first product, the TS‑210, targets Samsung Foundry’s 8 nm Low Power Ultimate (8LPU) process and is slated for a multi‑project wafer (MPW) shuttle by year‑end. In practice, TriSilica enables on‑device inference for vision, audio, and radar workloads while staying under a few hundred milliwatts, a sweet spot for battery‑operated edge devices.
Why the partnership matters
FotoNation brings a mature perception‑fusion IP stack that has already powered several niche cameras and smart‑sensor solutions. SEMIFIVE contributes deep expertise in high‑complexity, low‑power IC design and advanced packaging—capabilities that were decisive in winning the project. Together, they promise a faster time‑to‑market for edge AI chips that can run sophisticated models without relying on cloud connectivity.
Competitive landscape
The edge AI chip market is crowded. NVIDIA’s Jetson series, Google’s Edge TPU, and Qualcomm’s Snapdragon AI Engine dominate the mainstream, but they often trade power efficiency for raw performance. In contrast, TriSilica’s design philosophy mirrors the emerging “sensor‑first” paradigm championed by Apple’s Neural Engine and Amazon’s Graviton‑based inference chips, where multimodal data is processed locally to reduce latency and bandwidth costs. According to IDC, low‑power AI silicon for edge devices is projected to grow at a 30 % compound annual growth rate (CAGR) through 2027, outpacing general‑purpose AI accelerators.
Implications for enterprise AI
Enterprises that deploy AI at the edge—retail stores, manufacturing floors, autonomous drones, and smart‑city sensors—stand to benefit from TriSilica’s power envelope. By keeping inference on‑device, organizations can lower data‑transfer fees, meet strict privacy regulations, and achieve sub‑10 ms response times required for real‑time decision making. Marketing teams can embed AI‑driven visual analytics into point‑of‑sale displays or digital signage without a costly infrastructure overhaul, enabling personalized content delivery that reacts instantly to foot traffic patterns.
Looking ahead
If the TS‑210 validates FotoNation’s vision, the roadmap envisions a family of chips scaling from sub‑100 mW to sub‑1 W while supporting larger transformer‑based models. Such scalability could attract enterprise customers already entrenched in ecosystems like Microsoft Azure IoT, Salesforce Einstein, or Adobe Experience Platform, where on‑premise AI inference is increasingly demanded.
Market Landscape
The global edge AI hardware market is projected by Gartner to exceed $25 billion by 2027, driven by the need for real‑time analytics and data‑privacy compliance. While large‑scale cloud AI platforms from Google, Amazon, and Microsoft dominate centralized workloads, the shift toward decentralized processing is accelerating. Low‑power chips like TriSilica address a growing segment of the market that requires continuous operation on battery‑powered devices—think autonomous drones, wearables, and smart‑city sensors. Forrester estimates that enterprises will allocate 15 % of their AI spend to edge solutions within the next two years, underscoring the commercial relevance of this partnership.
Top Insights
- TriSilica’s multimodal sensor fusion differentiates it from vision‑only edge chips, opening new use cases in automotive radar and industrial spectroscopy.
- SEMIFIVE’s 8 nm LPU process delivers a 40 % power advantage over competing 14 nm solutions, crucial for battery‑constrained deployments.
- The partnership gives FotoNation a direct pathway to European OEMs, while SEMIFIVE gains a foothold in the EU’s emerging AI‑chip ecosystem.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
