As #intelligenttransportationsystems continue to evolve, #edgeAI is becoming the cornerstone of safer and smarter mobility. From real-time #trafficmonitoring to in-vehicle #videoanalytics and mobile SD-WAN connectivity, today’s transportation infrastructure demands rugged, intelligent computing at the edge. Lanner’s industry-certified Edge AI platforms are purpose-built to meet these demands, integrating high-performance AI acceleration powered by NVIDIA and Intel, along with secure cellular bonding for reliable data transmission from edge to cloud. https://lnkd.in/gc-FPmCC

The DGX Spark represents a significant evolution in the compute landscape, marking a strategic shift for Nvidia. Developers have traditionally depended on data centers and cloud GPUs. However, with 1 petaflop of local compute in a 2.6-lb device, Nvidia is redefining the line between enterprise-scale AI and personal development. This innovation goes beyond just speed; it emphasizes workflow control: - Fine-tune up to 70B-parameter models locally - Prototype offline and deploy via DGX Cloud - Reduce reliance on external infrastructure The DGX Spark signifies a hybrid future, where AI iteration cycles can be dramatically shortened, and hardware once again becomes a key competitive differentiator. #NVIDIA #AI #DeepLearning #EdgeAI #ComputeInfrastructure

From Cloud to Edge: how to transform your business model with Edge AI? 🤔 Our founder, Hervé Bath, was at #SIDO2025 to share with demos how cutting-edge AI meets embedded systems👇 Edge AI is the sweet spot for many enterprise challenges: minimal power consumption, low latency, and enhancing GDPR compliance through local processing. We teamed up with our partner STMicroelectronics to demonstrate three computer vision demos running entirely on the STM32N6 microcontroller. No cloud. Just a small, affordable chip doing AI work! 𝗧𝗵𝗲 𝘁𝗵𝗿𝗲𝗲 𝗱𝗲𝗺𝗼𝘀 🔒 Real-time Face Blurring: privacy protection happening directly on-device.  👁️ Object Detection: classic computer vision, but running locally with speed and accuracy.  🌈 Depth Estimation: the most complex of the three: generating full depth maps with heat-map visualisation. Check out the full demo in Hervé's interview with ST! 🎬https://lnkd.in/eviGDwug Curious about Edge AI for your use case? Let's explore the possibilities together! 📩 #euranova #enxproud #craft #STPartnerProgram

Ever wondered how to run a 60 GB+ LLM on a $1,600 RTX 4090? Huawei’s Computing Systems Lab just dropped a game‑changing open‑source method that shrinks models by 60‑70 % with the new SINQ quantization—calibration‑free and ~30× faster than AWQ. The result? Inference on consumer GPUs, cutting cloud costs by $3–4.50 per hour and democratising access to advanced AI. For Europe, this could mean less reliance on costly enterprise GPUs from a handful of vendors and a step toward true tech sovereignty—local hardware can power high‑performance LLMs without breaking the bank. What are your thoughts on adopting Huawei‑based compression for your AI stack? Will it shift the balance of cloud‑provider dominance, or raise new security discussions? Read the full article here: https://lnkd.in/eGivPjZc 🧠💻🔧🌍 #AI #Technology #OpenSource #TechSovereignty #LLM

$ARM Valued at $163.6B, ARM Holdings, LP dominates global chip architecture as AI and cloud demand skyrocket in 2025. https://lnkd.in/g2PmepwR

Shakti LLM × Qualcomm NPU We’re excited to showcase Shakti 2.5B, our sovereign small language model - optimized for Edge and low-power environments - now running on Qualcomm’s X-Elite silicon via the Qualcomm Developer Cloud (QDC). The live RAG demo, fully on-device, generates 18–22 tokens per second, showcasing the power of Shakti LLMs on Qualcomm NPU. Captured at the Qualcomm booth, India Mobile Congress 2025, this demo highlights the future of sovereign on-device AI. 🔗 Learn more: https://llms.sandlogic.com #ShaktiLLM #Qualcomm #QDC #XElite #RAG #EdgeAI #OnDeviceAI #IMC2025 #SovereignAI #SandLogic #MakeInIndia

"DGX Spark will usher in the next era of AI prototyping and advance our mission of making great technology more accessible from the edge to the cloud, while solving performance and power consumption challenges," said Vince Hu Vice President of MediaTek's Data Center and Compute Business Group. "The GB10 Superchip utilizes our high-performance computing expertise for the data center in combination with our power savings technologies for consumer devices, custom-built to run AI workloads." https://lnkd.in/daKKYkcJ

💾 The AI data center race is heating up - and one player is pulling far ahead. By 2027, NVIDIA’s data center sales are expected to soar past 300B, while AMD is projected to reach over 27B. From powering AI models to fueling entire cloud infrastructures, NVIDIA’s dominance shows just how central GPUs have become to the digital economy. ⚡️ This is not an offer, solicitation or investment advice. 

Qualcomm and MediaTek are expanding their rivalry into the cloud AI ASIC market. With smartphone sales slowing, both companies are focusing on cloud computing chips as a new growth area. They announced their strategies at COMPUTEX 2025 and plan to join NVIDIA's NVLink Fusion consortium. MediaTek is leveraging speed and partnerships, while Qualcomm is focusing on integration and technology depth. Read more: https://lnkd.in/eqe_aTDZ 📰 Subscribe to the weekly Silicon Brief Newsletter: https://lnkd.in/ejfzg92J #ai #artificialintelligence #ainews #aichips #datacenters