Gasgoo Munich- Black Sesame Technologies has signed a strategic cooperation agreement with Sharetronic Data Technology Co., Ltd. (SHARETRONIC) to jointly develop AI inference products and services, according to post Black Sesame Technologies issued on September 22 via its WeChat account.
Image source: Black Sesame Technologies
Key Takeaways
·The companies plan to combine computing infrastructure, complete-system production and AI inference services in an integrated solution, while exploring custom chips for next-generation inference applications.
·The partnership will combine Black Sesame Technologies' chip and software capabilities with SHARETRONIC's server-product, supply-chain and customer-service resources.
·Black Sesame Technologies will provide inference hardware based on its internally developed AI computing chips, together with a full software stack, development tools and technical support.
·SHARETRONIC will contribute its supply-chain, distribution-channel, computing-server and customer-service capabilities, while the two companies jointly pursue customers and optimize the proposed solutions.
Why It Matters
According to the announcement, demand for AI inference computing is expanding beyond the cloud into enterprise, industrial, robotics and intelligent-device applications. The partnership therefore targets factors such as inference throughput, cost per generated token, energy efficiency and hardware-software coordination, rather than focusing solely on peak computing performance.
For potential customers, the combination could provide a more integrated route from inference chips and software tools to finished servers and computing services. Whether that integration produces measurable cost or efficiency improvements has not yet been disclosed.
Background
Black Sesame Technologies has developed an AI inference server built around a multi-node interconnected system using high-performance chips. The company says the server supports local deployment and operation of large AI models and is compatible with high-performance models from leading AI companies. It is intended for users including small and medium-sized businesses, one-person companies and individuals.
Data and Evidence
The announcement cites an industry forecast that the AI inference computing market across data centers, edge systems and end devices will reach the trillion-dollar level by 2030. The source does not identify the organization behind the forecast or specify its methodology and market definition.
What to Watch Next
The companies plan to explore customized AI inference chips based on the computing requirements of industry customers. The potential scope covers chip specifications, architecture and intellectual-property design, tape-out, packaging and testing, as well as supporting software stacks and development tools. No development schedule, customer program or production target was disclosed.
