SemiKong, a new LLM trained by Aitomatic and its partners in the "AI Alliance", is the world's first large language model specifically crafted to serve the semiconductor industry's needs. SemiKong seeks to become part of the workflows of semiconductor design companies and act as a digital expert in the field, contributing to getting new chips out the door significantly faster.
According to Aitomatic, the company responsible for developing the SemiKong LLM, the semiconductor industry desperately needs to collect expert information. Many aging experts are retiring and taking their knowledge with them, so several companies are suffering from major expertise gaps. A purpose-built LLM trained for the industry's needs seemed like a solid way to give newer engineers the information they need to stay competitive.
Based on Meta's Llama 3.1 LLM platform, SemiKong recently saw the release of its 70B variant. Aitomatic and other partners in the new AI Alliance (including Meta, AMD, and IBM) collaborated on the LLM, with Aitomatic's DXA system becoming the backbone of SemiKong's deployment.
DXAs, or Domain-Expert Agents, are Aitomatic's way of connecting smaller LLM agents with the central hivemind of SemiKong 70B. By training on a client company's technical libraries or entries from expert engineers, DXAs can be keyed to that company's needs. The trained DXAs are then utilized by a core SemiKong deployment to automate development tasks or provide chatbot-style communication with engineers and workers.
In its current 70B form, and with smaller SemiKong-based DXA agents, the LLM has far outpaced the usefulness of generalized AI models in the semiconductor field. SemiKong advertises a 20-30% reduction in time to market for new chip designs and a 20% improvement in first-time-right manufacturing scores. It also claims to speed up the learning curves of new engineers by up to 50%, a major claim supported by Meta.
Those interested in deploying SemiKong's 70B model can download it from its website. SemiKong is one of the first projects to result from collaborations within the nascent AI Alliance, announced in December 2023. One of many new corporate alliances seemingly built to counter Nvidia's dominance in the tech industry, the AI Alliance includes major corporations like IBM and AMD and research institutions such as Yale and the University of Tokyo.
