AI computing is on the rise, and beyond the fundamental hardware components like CPU, GPU, and NPU, the crucial aspect lies in the ecosystem integration of application software and scenarios. Only by truly transforming people's daily work and life experiences through applications, bringing genuine convenience, can AI computing achieve success.
As a pioneer in the industry, Intel not only spearheaded the concept of AI computing but also introduced the groundbreaking Core Ultra processors and a plethora of AI-powered laptops, leading PCs into the AI era comprehensively. Intel has spared no effort in expanding the ecosystem, solidifying its position as an industry leader.
According to Intel's grand vision, the shipment volume of AI PCs in 2024 is expected to reach around 40 million units, and by 2025, this market size will exceed 100 million units, entering millions of households.
In order to enable developers to better utilize various AI tools, hardware, and software, Intel has fully leveraged its comprehensive accumulation in open source, technology, and ecosystem over the past 20 years to drive the rapid development of AIGC.
Currently, globally, there are over 20 categories of 500 AI models optimized on Intel Core Ultra, covering various categories of local AI inference, including natural language processing, diffusion processing, super-resolution, object detection, image classification/segmentation, computer vision, and more, such as OpenVINO Model Zoo, Hugging Face, ONNX Model Zoo, PyTorch, Phi-2, Mistral, Llama, Bert, Whisper, Stable Diffusion 1.5, and so on.
On December 16, 2023, just one day after the official release of the Core Ultra processor, Intel launched the AI Innovation Challenge, providing a platform for developers to showcase their creativity and achievements, helping each developer seize the opportunities brought by generative AI.
In the end, outstanding works from 30 teams stood out and advanced to the final competition. The VidBot team from the individual track and the Panda AI team from the enterprise track both won first prizes, each receiving a prize of 50,000 RMB.
Additionally, there were 2 second prizes, 3 third prizes, and 9 special prizes for each track, with a total prize pool of 525,000 RMB.
The competition is divided into individual and enterprise tracks, with a total of up to 2120 teams registering and submitting 1083 entries.
These AI applications are innovative, creative, practical, and deployable, covering a wide range of fields such as the internet, education, healthcare, agriculture, construction, law, retail, culture and tourism, industry, energy, gaming, film and television, advertising, and more.
This clearly demonstrates the enthusiasm for generative AI and the high expectations and optimism for its broad prospects.
During the preliminary stage in February this year, the competition held 10 training and community coaching sessions. In March, there was an innovation application special competition, followed by expert coaching and in-depth optimization stages in April and May.
To assist participants and developers, Intel not only offered its own open-source software frameworks and optimization toolchains like OpenVINO, oneAPI, and IPEX-LLM, but also collaborated with the entire ecosystem to provide a wealth of AI large models covering various aspects of generative AI.
Interestingly, Intel also disclosed the usage of XPU in their entry.
CPUs are suitable for general workloads, offering fast response times and low latency. GPUs provide high computing power and throughput, significantly accelerating content creation. NPUs, on the other hand, are designed for sustained low workloads with the lowest power consumption. Each of the three engines has its own characteristics, working together to form Intel's XPU strategy.
Currently, CPU+GPU remains the absolute mainstream, with a much higher usage rate compared to other solutions. Encouragingly, new applications that utilize all three engines - CPU, GPU, and NPU - are continuously emerging.
As the exclusive AI PC partner for the competition, Lenovo has provided strong support, including AI PC devices, application development and ecosystem, prizes, and professional judging.
At the same time, Lenovo is comprehensively advancing AI PC products and applications, particularly pioneering the definition of five key features of AI PCs: embedded personal large models for natural user interaction, local heterogeneous AI computing power (CPU/GPU/NPU), personal knowledge base, open AI application ecosystem, and personal data and privacy protection.
In terms of the ecosystem, Lenovo's AI PCs come pre-installed with a comprehensive range of AI applications, with the first batch including ten applications covering various experiences, such as AI artist, AI PPT, document summarization, knowledge Q&A, AI image recognition, meeting minutes, intelligent assistant, device optimization, AI device migration, and AI services.
The recent AI Innovation Application Competition may have come to a close, but it is far from over. Intel will continue to nurture the competition's projects through various efforts such as ecosystem channels, marketing promotions, and technical support, bringing AI from the cloud to benefit everyone.
More importantly, AI on PCs is just getting started, and Intel's initiatives will unfold one after another. They will persist in supporting developers to enter the new AI market through training, application promotion, resource matching, and market expansion. The previously launched AI PC Developer Program and Enterprise AI Open Platform will also continue to strengthen, promising a bright future.
Now, let's take a look at the brief introductions of the outstanding works from the AI Innovation Application Competition:
