“Edge computing and machine vision technologies are quite mature, and more attention and resources may be devoted to the development and application of large language models and AI in the future.” AI models may gradually move from the cloud to the edge. We can expect a huge wave of innovation to be brewing. There are already a number of mature solutions on the market and we are at a critical point in this transition.” During the 2024 Intel Network and Edge Computing Industry Conference, Sachin Katti, senior vice president and general manager of the network and Edge Division of Intel Corporation, said in an interview with Caixin.

ANB10D-425CU2N Recently, Cailiannews reporters in the industry learned that from the computing power, edge reasoning ability, cost comprehensive considerations, the penetration of artificial intelligence in the edge field will become a key trend, a number of giants into the track, many mature solutions emerge, edge AI is on the rise, and the global edge market size will reach 100 billion dollars in 2030.

Mature solutions for AI aimed at the edge emerge

Cailiannews reporter recently learned from Intel that in the first half of this year, Intel China’s edge business maintained year-on-year growth, and the company expects this trend to continue. Edge AI has been a focus for Intel over the past few years, with deployments in retail, medicine, logistics, agriculture, transportation, manufacturing, and more.

Sachin Katti told CaiAP reporter, “Now we talk about edge AI has actually progressed from edge machine vision to such as large language models, edge applications and generative AI, so Intel will continue to provide relevant capabilities to accelerate the deployment of generative AI and large language models at the edge.” “We’ve expanded beyond the hardware to the application layer, software and customer service support.”

ANB10D-425CU2N

During this year’s MWC, Intel said it has completed more than 90,000 actual deployments at the edge. Sachin Katti said at the conference that Intel will take an open, modular, unified platform as the core of the approach to accelerate the deployment of edge Al solutions. “On the hardware side, we will provide GPU and NPU solutions with enough memory, and many models such as Llama 2 and Llama 3 can run locally.” In terms of ecology, Intel’s Network and Edge division has established cooperative relationships with more than 500 Oems/ODMs and more than 150 ISVs in China.

In fact, not only Intel, AMD, Nvidia, Qualcomm and other giants, Yuntian Lifei (688343.SH), China Science and Technology Chuangda (300496.SZ), Guoke Micro (300672.SZ) and other A-share listed companies have been aimed at the edge of the action.

ANB10D-425CU2N At the same time, Caixin reporters noticed in the exhibition area of the conference that A-share manufacturers such as robots (300024.SZ), Deep convincing (300454.SZ), digital video (300079.SZ), and China Science and Technology Chuang Da all displayed their own edge AI achievements.

In the exhibition area, the reporter of Caiyun News noticed a case of data annotation through automated means in the process of model training. Booth staff told reporters that based on Intel core processor, Intel Sharp independent graphics card ARC770 to create the CNN edge training program, with the help of visual large model to achieve automatic picture annotation and MLFlow based model automatic retraining. “The original training model to find the algorithm company, the annotation needs people to complete, now you can use the machine annotation, training costs, labor costs are greatly reduced, and now the industry is slowly appearing this kind of training and promotion program.”

Intel’s newly created 3D virtual Homo sapiens “Xiao Ying” is also on display. Booth staff said that “Xiao Ying” is also based on core processors, ARC770 graphics card, fully deployed on the edge, but also locally integrated large language model and RAG technology. It is reported that it can be competent for business hall customer service, shopping mall guide, museum docent and other roles.