Kog aims to boost LLM inference speed on existing GPUs

Key facts

• Kog is a French startup developing AI inference acceleration software.
• The software targets large language models (LLMs).
• The technology aims to improve inference speed on standard datacenter GPUs.
• Kog was founded by a former cybersecurity expert.
• The company aims for 10x speed improvements.
• This speed improvement target is set for September.
• The company is seeking Series A funding.

French startup Kog is working on software designed to accelerate artificial intelligence inference, specifically for large language models (LLMs), on existing datacenter GPUs. The company's technology aims to provide substantial speed enhancements compared to current inference methods. Kog was founded by an individual with a background in cybersecurity. The startup's immediate goal is to demonstrate a tenfold improvement in inference speed by September. This demonstration is intended to help Kog secure Series A funding.