Kneron, a San Diego-based full-stack AI company known for pioneering neural processing units (NPUs), has released its latest edge AI servers and AI embedded PCs. To date, Kneron has raised approximately $200 million, backed by Horizons Ventures, Qualcomm Ventures, Sequoia, Foxconn, and others.
The KNEO 330 is Kneron's newest and second private edge GPT server. The first unit launched in 2023, the KNEO 300, already has enterprise customers in manufacturing, financial services, and universities such as Stanford and UCLA. The KNEO 330 boasts 48 TOPS of AI computing power, up to 8 simultaneous connections, LLM support, and stable spreading. Its RAG accuracy is remarkably similar to cloud solutions under much lower hardware conditions. The KNEO 330 significantly reduces the total cost of AI for small and medium-sized businesses by 30%-40%.
AI ML News:
Compared to most cloud solutions, KNEO 330 facilitates integration and hierarchical permission management, significantly improving privacy and security. Its comprehensive features include an offline version of multimodal GPT.
“We are pleased to partner with Kneron to bring our technology to market, and we look forward to working with them to bring our technology to market,” said Albert Liu, CEO and founder of Kneron. “AI has recently experienced a boom, and powerful AI models are advancing faster than many experts thought possible. Striking a balance between ethical AI and profit-driven AI is a real challenge. Concerns about the power and data required to train AI models, the possibility of AI hallucinations, and other issues are real. We believe our product will be the key to solving the current GPT ESG and power consumption issues.”
In addition to the KNEO 330, Kneron will release an AI embedded PC equipped with the third-generation NPU chip KL830. According to an IDC report, the era of AI PCs is upon us, with sales expected to increase from 50 million units in 2024 to more than 167 million units in 2027, accounting for more than 60% of the total PC market. Meanwhile, Gartner predicts that global shipments of AI PCs and AI smartphones will reach 295 million units in 2024, more than 10 times the 29 million units in 2023.
Featured AI startups: Agricultural technology company Grabmarket releases “AI ordering” for food supply chain market
The power and cost of the KL830 enables lower cost AI PCs, accelerating adoption among the broader consumer market. The KL830 provides up to 10eTOPS@8bit Consolidated Compute Power (CCP) with a peak power consumption of 2W. The NPU enables personalized GPT.
It demonstrates the power and efficiency of NPUs, and when paired with flagship GPUs, reduces energy usage by 30% and extends product lifespan, pointing to the future of NPUs in high-performance, yet affordable gaming PCs.
This chip is ready to use in AIoT devices. The KL830 chip maintains consistent fixed-point and floating-point precision. The KL830 is also available via a USB dongle that can edge AI-enable any device, such as broadband routers, IoT cameras, and traditional PCs. It offers 10eTOPS@8bit and supports many large language models with lower parameters.
Designed for developers, the KNEO Platform is an easy-to-use open marketplace for Edge GPT. Kneron is a user-friendly Deploying AI-generated content models It runs through Kneron's compiler using the “Hugging Face” link, and users can switch to Edge GPT if they wish.
Kneron's Edge GPT Service (EGaaS), consisting of a developer platform, management platform and edge GPT warehouse, enables devices to process data locally, enabling real-time decision-making without relying on the internet or cloud servers. This approach improves speed, supports multi-modal capabilities, and strengthens privacy and security. Kneron's comprehensive enterprise edge GPT solution can be customized to fit each company's needs by training and deploying large language models (LLMs) for a variety of scenarios.
During his keynote address at the Computex event in Taipei, Kneron will highlight the following points:
-
Our newest and second-largest private edge GPT server, the KNEO 330, prioritizes privacy and security while lowering costs for SMBs by 30-40%.
-
KL830, its edge GPT AI chip, works perfectly in AI PCs, USB dongles and edge servers, and its NPU can reduce energy consumption by 30% when combined with leading GPUs.
-
Additionally, Kneron discussed its fourth-generation NPU, the KL1140, which can be used for both edge and cloud AI applications and is expected to be available in 2025.
Kneron has raised nearly $200 million from Horizons Ventures, Qualcomm, Sequoia, Foxconn, Delta, Vivotek, and others. Kneron provides end-to-end integrated hardware and software solutions that enable on-device edge AI inference. Qualcomm, Toyota, Kenwood, Garmin, Panasonic, Quanta, Compal, Unimicro, MiTAC, Hanwha, Spark, Naver, TUL, and Gree have adopted Kneron solutions.
Founded in 2015 and based in San Diego, Kneron develops full-stack hardware and software products for AI applications. Kneron's lightweight, reconfigurable solutions solve the three biggest problems facing AI use cases: latency, security, and cost, enabling AI everywhere.
AI ML Insights: Employee data risks in an AI-driven world
