Termsslug: edge-computing

Edge Computing

边缘计算

A distributed computing paradigm that brings computation and data storage closer to the location where it is needed.

Curated Intelligence

28 items · Page 1/3
The Verge AIT2·Infra
78 pts
Nvidia Posts $115.8B Quarterly Revenue as Data Center Sales Double

Nvidia reported fiscal Q3 2027 revenue of $115.8 billion, up 103% year over year and above its $108 billion guidance, joining the small club of companies to top $100 billion in a single quarter. Data center revenue reached $107.6 billion, up 110%, while edge computing including consumer gaming brought in $8.2 billion, up 42%. Interim CEO Colette Kress did not initially address former CEO Jensen Huang's departure or a possible return, topics likely to surface on the analyst call.

InfoQ 中文T2·Infra
62 pts
Cloudflare Opens KV Instant with 1.62ms p99 Read Latency

Cloudflare has opened its internal Quicksilver configuration store to customers as Workers KV Instant, cutting p99 read latency from 287ms in classic KV to 1.62ms. The mode is tightly constrained: 10,000 keys and 1MB total per namespace, one write per second per namespace, and notably higher storage and write pricing, targeting single-key config workloads read globally. It remains in invite-only private beta.

IT之家 · IT业界T2·Industry
58 pts
Cloudflare Annualized Revenue Tops $3B as Q3 Growth Hits Four-Year High

Cloudflare reported Q3 FY2026 revenue of $782.5 million, up 39% year over year, its fastest quarterly growth in nearly four years, pushing annualized revenue past $3 billion for the first time. Non-GAAP operating margin rose to a record 17.5%, large-customer and Workers developer counts hit new highs, and guidance for Q4 and the full year came in ahead of prior expectations.

机器之心 (微信公众号)T2·Infra
68 pts
Iluvatar CoreX Launches Edge Compute Terminal for Local Robot Training

Iluvatar CoreX unveiled its edge compute terminal, Tongyang TY1200, at a Shanghai media day. The device independently completed full robot simulation training workflows locally, with measured performance surpassing NVIDIA's Thor T5000. The company also previewed the upcoming TY2000, targeting 600 TOPS of dense compute to bring cloud-grade AI capabilities to the edge.

The DecoderT2·Agents
68 pts
SETI Deploys AI Agents for Real-Time Telescope Control and Signal Analysis

The SETI Institute and Breakthrough Listen have deployed AI agents based on Nvidia's Holoscan platform at the Allen Telescope Array to control telescopes and analyze radio signals in real time. The system reduces data processing from weeks to hours and achieved the first real-time AI detection of a pulsar in the Crab Nebula, with plans to test open-source edge models for future lunar sensors.

IT之家 · IT业界T2·Infra
42 pts
Silicon Motion Hits Record Q3 Revenue Amid AI Storage Pivot

Silicon Motion reported record-breaking Q3 2026 revenue of $560 million, marking its fourth consecutive quarterly high. The company is pivoting from consumer NAND controllers to AI infrastructure and edge computing storage solutions, with enterprise and AI product lines showing significant early-stage growth. Management projects full-year revenue to double year-over-year.

Hugging Face BlogT1·Models
45 pts
LiquidAI Releases LFM3 for Custom Multimodal Edge Models

LiquidAI has introduced the LFM3 framework to enable developers to build specialized multimodal large models optimized for edge devices. The tool focuses on efficient inference and custom deployment in resource-constrained environments, currently in an early community sharing phase.

AWS Machine Learning BlogT1·Tools
45 pts
SageMaker SDK v3 Updates Edge CV Model Lifecycle Pattern

AWS updates the best practice for edge computer vision model lifecycle management using SageMaker Python SDK v3. The pattern addresses large-scale quality inspection scenarios where environmental variations require camera-specific fine-tuned models, replacing deprecated SDK workflows.

InfoQ 中文T2·Infra
68 pts
Cloudflare Launches K2 Serverless Event Streams on R2

Cloudflare has launched K2, a serverless event streaming service built on R2 object storage to replace traditional Kafka architectures. By using in-memory staging and batch writes for log append semantics, K2 trades ~1s p99 latency for cost efficiency and elasticity, specifically optimized for ephemeral compute and wide-area edge networks.