2025-11-26
Kneron Launches KL1140 Chip, Bringing High-Performance AI Directly to Devices
●First NPU chip to run full Mamba networks on-device
●Enables real-time AI applications at 10X cost reduction and 3X improvement in energy efficiency, versus existing cloud solutions
San Diego, California, November 26, 2025– Kneron, the San Diego-based full-stack AI company pioneering neural processing units (NPUs), today announced the launch of its KL1140 chip, marking a major milestone in AI computing. The KL1140 brings powerful LLMs to edge devices, delivering high-performance AI with up to 3X greater energy efficiency and a 10X reduction in cost compared to current solutions.
The KL1140 comes at a critical inflection point for the AI industry. As real-world applications of AI accelerate and trillions in capex are earmarked for data center infrastructure, the industry is at risk of collapsing under its own computational and energy demands.
AI companies are struggling to contain inference costs that continue to rise rather than fall, and global energy demand from data centers is projected to reach 175GW or more by 2035. Cloud-based AI is increasingly expensive, slow, power-hungry, and less secure.
“The twin threat of high costs and vast energy consumption means the status quo of AI computing is fundamentally unsustainable,” said Albert Liu, Founder and CEO of Kneron. “The KL1140 is our response to the challenges of scaling LLMs in the cloud alone. By running advanced models at the edge, we’re achieving a technical milestone that opens up entirely new applications for everyday devices, putting the power of LLMs directly into the hands of users.”
Breaking the Edge AI Performance Barrier
The KL1140 is the first NPU chip capable of running full Mamba networks at the edge, a technical milestone that moves powerful LLMs out of costly cloud data centers and into portable devices. Cascading four KL1140 chips can deliver performance equivalent to a GPU for running models with up to 120 billion parameters, while consuming just one-third to half the power and reducing hardware costs by 10x. Independent benchmarking by UC Berkeley has confirmed Kneron as the first edge processor to break the efficiency barrier.
Designed for real-time natural language processing, voice interfaces, intelligent vision, robotics, and more, the KL1140 enables developers and enterprises to deploy sophisticated AI applications locally and securely on portable devices without reliance on cloud infrastructure. It also removes the lag associated with cloud responses.
Real-world applications of the KL1140 can include:
●A security robot that understands natural language commands and recognizes complex situations—without needing a WiFi connection to a data center
●An automotive system that runs sophisticated AI for voice commands and decision-making entirely in the car – no cloud lag, works even without cell service
●A private enterprise AI assistant running on a small edge server in an office – keeping sensitive data on-premises instead of sending it to the cloud
●Smart manufacturing equipment that can analyze video, understand voice commands, and make intelligent decisions locally on the factory floor
“The arrival of the KL1140 is more than just another chip launch, it’s a tipping point in the journey towards practical, high-performance and sustainable AI,” said Liu. “By bringing intelligence to the edge, we’re enabling developers and enterprises to create applications that were impossible before.”
Kneron has rapidly expanded from an edge chip designer into a full-stack AI infrastructure company. It has already delivered sovereign AI projects for hospitals, universities, and government agencies, proving its ability to support secure, local AI deployments. Kneron is also growing its Edge AI ecosystem through its KNEO Pi developer platform, which is already used by more than 28,000 developers worldwide. In parallel, its partnership with Taiwan Spark Technology will enable the joint manufacture of LLM servers powered by Kneron chips. Together, these efforts position Kneron as one of the few companies building AI infrastructure end-to-end, from chips to server systems.
Since its establishment in 2015, Kneron has been recognized for its reconfigurable NPU architecture and has received awards, including the IEEE CAS Darlington Award for breakthrough technologies. The company supports customers across AIoT, security, automotive, and edge server applications, including Toyota, Quanta, Hanwha, and Dessmann, among others, helping drive innovation while reducing latency, energy use, and costs.
About Kneron
Founded in 2015 and based in San Diego, Kneron develops full-stack hardware and software products for AI applications. Kneron’s lightweight reconfigurable solutions resolve three major problems faced by AI use cases—latency, security, and cost–thereby enabling AI everywhere. To date, Kneron has raised over $200 million, backed by Horizons Ventures, Qualcomm Ventures, Sequoia, Foxconn, and more. For further information about Kneron, please visit: http://www.kneron.com/about.php
●Enables real-time AI applications at 10X cost reduction and 3X improvement in energy efficiency, versus existing cloud solutions
San Diego, California, November 26, 2025– Kneron, the San Diego-based full-stack AI company pioneering neural processing units (NPUs), today announced the launch of its KL1140 chip, marking a major milestone in AI computing. The KL1140 brings powerful LLMs to edge devices, delivering high-performance AI with up to 3X greater energy efficiency and a 10X reduction in cost compared to current solutions.
The KL1140 comes at a critical inflection point for the AI industry. As real-world applications of AI accelerate and trillions in capex are earmarked for data center infrastructure, the industry is at risk of collapsing under its own computational and energy demands.
AI companies are struggling to contain inference costs that continue to rise rather than fall, and global energy demand from data centers is projected to reach 175GW or more by 2035. Cloud-based AI is increasingly expensive, slow, power-hungry, and less secure.
“The twin threat of high costs and vast energy consumption means the status quo of AI computing is fundamentally unsustainable,” said Albert Liu, Founder and CEO of Kneron. “The KL1140 is our response to the challenges of scaling LLMs in the cloud alone. By running advanced models at the edge, we’re achieving a technical milestone that opens up entirely new applications for everyday devices, putting the power of LLMs directly into the hands of users.”
Breaking the Edge AI Performance Barrier
The KL1140 is the first NPU chip capable of running full Mamba networks at the edge, a technical milestone that moves powerful LLMs out of costly cloud data centers and into portable devices. Cascading four KL1140 chips can deliver performance equivalent to a GPU for running models with up to 120 billion parameters, while consuming just one-third to half the power and reducing hardware costs by 10x. Independent benchmarking by UC Berkeley has confirmed Kneron as the first edge processor to break the efficiency barrier.
Designed for real-time natural language processing, voice interfaces, intelligent vision, robotics, and more, the KL1140 enables developers and enterprises to deploy sophisticated AI applications locally and securely on portable devices without reliance on cloud infrastructure. It also removes the lag associated with cloud responses.
Real-world applications of the KL1140 can include:
●A security robot that understands natural language commands and recognizes complex situations—without needing a WiFi connection to a data center
●An automotive system that runs sophisticated AI for voice commands and decision-making entirely in the car – no cloud lag, works even without cell service
●A private enterprise AI assistant running on a small edge server in an office – keeping sensitive data on-premises instead of sending it to the cloud
●Smart manufacturing equipment that can analyze video, understand voice commands, and make intelligent decisions locally on the factory floor
“The arrival of the KL1140 is more than just another chip launch, it’s a tipping point in the journey towards practical, high-performance and sustainable AI,” said Liu. “By bringing intelligence to the edge, we’re enabling developers and enterprises to create applications that were impossible before.”
Kneron has rapidly expanded from an edge chip designer into a full-stack AI infrastructure company. It has already delivered sovereign AI projects for hospitals, universities, and government agencies, proving its ability to support secure, local AI deployments. Kneron is also growing its Edge AI ecosystem through its KNEO Pi developer platform, which is already used by more than 28,000 developers worldwide. In parallel, its partnership with Taiwan Spark Technology will enable the joint manufacture of LLM servers powered by Kneron chips. Together, these efforts position Kneron as one of the few companies building AI infrastructure end-to-end, from chips to server systems.
Since its establishment in 2015, Kneron has been recognized for its reconfigurable NPU architecture and has received awards, including the IEEE CAS Darlington Award for breakthrough technologies. The company supports customers across AIoT, security, automotive, and edge server applications, including Toyota, Quanta, Hanwha, and Dessmann, among others, helping drive innovation while reducing latency, energy use, and costs.
About Kneron
Founded in 2015 and based in San Diego, Kneron develops full-stack hardware and software products for AI applications. Kneron’s lightweight reconfigurable solutions resolve three major problems faced by AI use cases—latency, security, and cost–thereby enabling AI everywhere. To date, Kneron has raised over $200 million, backed by Horizons Ventures, Qualcomm Ventures, Sequoia, Foxconn, and more. For further information about Kneron, please visit: http://www.kneron.com/about.php
