AI Hardware Innovations and Software Transformations in 2024

In 2024, AI hardware innovation reached new heights. The AI industry consensus became ‘focus on applications, not parameters.’ On one hand, hardware companies stacked AI capabilities, leading to an explosion in AI hardware across various sectors such as mobile phones, PCs, home appliances, automobiles, cleaning, smart homes, headphones, cameras, and storage. On the other hand, AI deeply transformed software, with native AI applications like Wen Xiaoyan and Doubao becoming widespread, and search engines, input methods, browsers, payment systems, office software, and e-commerce being reconstructed by AI.


AI has become the ‘magic wand’ of technological innovation, and this is just the beginning. In December, the ‘Lei Technology Annual’ special topic was launched, which systematically sorted out the companies, products, technologies, and figures in the technology industry worth recording in 2024, and ‘Looking Forward to 2025’ will ‘spoiler’ the products and technologies in the technology industry worth looking forward to in 2025, paying tribute to innovation, recording the times, and enlightening the future.


Please subscribe and follow. AI-modified videos continue to go viral, and Sora (Turbo) is open for use. Despite the progress bar of 2024 being about to end, new creative ideas and products continue to emerge in the technology industry. It can be said that the technology industry has delivered a very impressive answer in 2024. At the beginning of the year, the OPPO Find X7 series took the lead in AI phones, and Honor’s MagicOS 9.


0 ushered mobile phones into the ‘autonomous driving’ era. From the ‘pregnancy’ of Sora at the beginning of the year to the end of the year, AI videos have undergone earth-shaking changes. These seemingly unrelated breakthroughs together form the technology panorama of 2024 and are also defining the rules of the next era. Against this backdrop, there are always some technology products destined to be extraordinary.


They not only bear the significance of the development of technology but also promote changes in lifestyle to varying degrees. Next, let’s review these epoch-making technology products and see how they write the legend of this year. (The ranking is not in order). OpenAI o1 Model: The Starting Point of AI Inference Era Epoch-making: Shifting the focus of large models to reasoning ability, it opens up a new direction for AI evolution based on reinforcement learning.


On September 12, OpenAI launched the o1 model, code-named ‘Strawberry’. The core highlight of this new generation of large models is the introduction of a training architecture of ‘reinforcement learning + chain of thought’, attempting to solve the long-standing shortcomings of generative AI in complex reasoning. Compared to the previous GPT series models, the o1 model can more systematically break down complex problems into logically clear steps, providing answers closer to human thinking – whether in solving mathematical problems or logical reasoning, it has shown excellent ability.


The release of o1 by OpenAI has drawn intense attention from the outside world. The industry generally regards it as a major breakthrough in AI reasoning capabilities. Yang Zhilin, the founder of DarkSide, pointed out in a speech that the o1 model marks a new paradigm for general artificial intelligence (AGI). “It is achieved through reinforcement learning.” Undoubtedly, the emergence of the o1 model is epoch-making.


After the launch of o1, global large models quickly shifted from simply pursuing generative capabilities to reasoning capabilities driven by reinforcement learning. Manufacturers such as Tongyi of Alibaba, Kimi of DarkSide, and DeepSeek have also started “reinforcement learning” to explore the upper limits of AI capabilities in complex tasks. Without a doubt, the significance of the o1 model goes beyond a single product.


It indicates the next development direction of generative AI – intelligence no longer stops at “generation” but moves towards deeper “learning” and “reasoning”.



HarmonyOS NEXT: The Declaration of Independence of Domestic Operating Systems


Epoch-making: Completely getting rid of the dependence on Android, marking that China’s operating systems have entered a brand-new stage. On October 22, Huawei officially released the brand-new HarmonyOS NEXT (Hongmeng OS) operating system for the Internet of Everything. It adopts a full-stack self-developed architecture, including completely replacing the Linux kernel with the Hongmeng kernel and no longer being compatible with Android applications. Although there are still many aspects of HarmonyOS NEXT that need to be improved and iterated, it has undoubtedly demonstrated unique experience advantages.


On the other hand, the Hongmeng ecosystem has made rapid progress in the past year or so. As of the release, the official announced that more than 15,000 applications and meta-services have been launched on the Hongmeng application market. At the same time, the cross-platform capabilities and open ecosystem strategy of HarmonyOS NEXT are also attracting more developers to join and enrich the application supply. The epoch-making significance of HarmonyOS NEXT not only lies in getting rid of the dependence on Android and achieving autonomy and control. As a brand-new operating system, it is actually providing a brand-new ecosystem for the Internet of Everything, bringing new possibilities to the Chinese and global operating system markets.


Honor MagicOS 9.0: AI-Driven All-Scenario Smart OS


Epoch-making: Through the system-level AI – YOYO intelligent agent on MagicOS 9.0, it ushers in the era of “autopilot” for mobile phones.


On October 23rd, Honor officially released the MagicOS 9.0 powered by the YOYO AI assistant, featuring a brand-new four-layer AI architecture that deploys large language models, image models, voice models, and multimodal models with different parameter scales on various devices.


The most significant change comes from the YOYO AI assistant, which not only possesses core capabilities such as natural semantic understanding, user behavior learning, intent recognition, and decision-making but also plans complex tasks and performs autonomous operations on smartphones based on pure AI vision, completing complex tasks like ordering coffee with a single sentence.


The innovations in AI capabilities and user interaction of MagicOS 9.0 mark a new stage in the intelligence of mobile operating systems. Particularly, the combination of AI operating systems and system-level AI has transformed smartphones from passive tools to proactive assistants, indicating the direction for the future development of operating systems and propelling the transformation of the smartphone industry.


OPPO Find X7: The Tipping Point of the AI Smartphone Wave


A groundbreaking move: bringing AI from concept to practice, directly pushing the smartphone industry into the ‘AI smartphone’ era. On January 8th, OPPO launched the world’s first AI smartphone with a 7 billion parameter large model deployed on the device side—the Find X7 series—and soon officially announced the entry into the AI smartphone era.


With the arrival of ColorOS 14 (AI Spring Edition), new interactive experiences are offered, whether it’s the Xiaobu assistant supporting voice recognition and text generation, or AI call summaries and AIGC noise cancellation features.


The launch of the OPPO Find X7 series and the declaration of entering the AI smartphone era have ignited external expectations for AI smartphones from the start of the year and shattered the last layer of concerns.


On the other hand, the truly epoch-making significance of the Find X7 series lies in its influence on the mainstream mobile phone manufacturers’ direction. After OPPO’s release of the Find X7, major manufacturers have followed suit, launching ‘AI smartphones’ and increasingly aligning their functional implementations with the Find X7.


From AI call recordings to AI photo editing, from Honor to Xiaomi, and then to Apple, the smartphone industry has quickly sparked an ‘AI competition.’ In short, the Find X7 is not only a successful product but also a signal: AI technology is no longer a luxury but an indispensable core competitiveness of smartphones.


From concept to practice, Find X7 has indeed set a direction for the industry to a certain extent, defining the ‘AI standard’ for smartphones.


Sora: A Watershed Moment in AI Video Technology: After its release, it completely rewrote the technological path of AI video generation, propelling the industry to fully embrace the architecture design that integrates Diffusion and Transform. On February 16, OpenAI unveiled Sora for the first time, a generative model capable of transforming text directly into video, which was not officially opened to the public until December. Regardless, Sora is a phenomenon-level product that sparked intense global discussions and shook the AI industry at the beginning of the year. Sora is a significant milestone in the field of AI video generation, but what truly makes Sora epoch-making is the technological direction behind it.


Images, compressed and edited, are credited to / OpenAI. Sora employs a DiT-like (Diffusion + Transformer) architecture, completing video generation through diffusion models while introducing Transformer to enhance the correlation between text and visuals. Users only need to input a simple text description or provide reference materials, and Sora can generate videos with a maximum resolution of 1080p and a maximum duration of 20 seconds.


More importantly, the new architecture significantly improves the quality and semantic consistency of video generation, quickly becoming the industry standard. After the release of Sora, almost all mainstream technology companies began to shift towards the DiT-like architecture. Major models such as Kuaishou’s Keling, Google’s Veo, ByteDance’s Jimeng, and Tencent’s Hunyuan quickly followed suit, developing video generation models by integrating Diffusion and Transformer.


The significance of Sora lies not only in its own technological achievements but also in its complete definition of the technical trajectory of video generation models, allowing AI-generated videos to iterate at a speed that is almost ‘lightning-fast’. At the same time, AI video tools have taken shape, becoming increasingly capable of participating in video production processes.



Yushe G1: The Hardware Pioneer of Humanoid Robots, a Watershed Moment: By compressing the price below 100,000 yuan, it has rewritten the market pattern of humanoid robots with breakthroughs in hardware technology. On August 21, Yushe Technology officially launched the mass-produced version of the G1 humanoid robot, standing at about 127 centimeters tall and weighing 35 kilograms.


The Yu Shi G1 mass production version utilizes self-developed joint motors combined with optimized force-position hybrid control technology, which endows it with exceptional dynamic balance and flexibility. It is capable of performing complex movements such as trotting and stick dancing, with a payload capacity of 3 kilograms in its arms, and can even accurately complete delicate tasks like opening bottles and welding.


The Yu Shi G1 mass production version has impressed many with its humanoid robot hardware. NVIDIA scientists have reminded their peers on the importance of hardware technology in the field of robotics, highlighting that while Silicon Valley has long focused on software and AI, Yu Shi’s breakthroughs in hardware are equally far-reaching. What is even more eye-catching about the Yu Shi G1 mass production version is its price.


With a pricing of 99,000 yuan, it becomes the first humanoid robot to have its price controlled within 100,000 yuan, bringing it closer to the affordability range of consumers and small and medium-sized enterprises. The release of Yu Shi G1 is a milestone in robot hardware technology, proving to the world that high-performance robots do not necessarily require exorbitant costs, and it is changing the rules of the humanoid robot market.



‘Black Myth: Wukong’: An Ambitious 3A Game from China


Epoch-making: With its ultimate content and market performance, it has reshaped the global perception of Chinese 3A games and inspired confidence and ambition in Chinese 3A games. On August 20th, ‘Black Myth: Wukong’, developed by Game Science, was officially released, causing an internet sensation. Based on ‘Journey to the West’, ‘Black Myth: Wukong’ plays the ‘Destined One’ embarking on an adventure to find the six roots of Sun Wukong. The game’s combat system is refreshing, and its visual performance is on par with blockbuster films, with each scene seemingly narrating an epic of China.


After its release, ‘Black Myth: Wukong’ quickly rose to the top of global game sales charts. In just 3 days, sales exceeded 10 million copies; a month later, this number doubled to 20 million copies. The ‘Overwhelmingly Positive’ reviews on the Steam platform and the fervent pursuit by global players have made this domestic game a phenomenon. However, the significance of ‘Black Myth: Wukong’ goes far beyond commercial success.


It redefines the position of domestic games in the global market, prompting the world to take seriously the creativity of Chinese game developers. More importantly, this game has ignited the belief of domestic game developers in creating 3A masterpieces. Both giants like NetEase and Tencent and single-player game projects like ‘Shadow Blade Zero’ have begun to re-evaluate themselves.
Baidu Wenku, Zhi Xiaobao, and Doubao: Making AI an Integral Part of Life



Epoch-making: Through content creation, life services, and multi-hardware access, AI is redefining its integration into people’s daily lives. If 2023 was the year when AI assistants were collectively born, then 2024 can be considered the pivotal year for the popularization of AI assistants. From Baidu Wenku to Alipay, and Doubao by ByteDance, various AI products are exploring how to truly embed AI into people’s lives.


Baidu Wenku’s Comprehensive Upgrade: One-stop AI Content Acquisition and Creation Platform


At 15 years old, Baidu Wenku has made a leap in 2024 from a document resource platform to a ‘one-stop AI content acquisition and creation platform’. This year, Baidu Wenku has undergone earth-shaking changes, successively launching intelligent document generation, PPT production, and educational content creation functions, and integrating with Baidu Cloud, reducing the threshold for content creation to the lowest. The AI transformation of Baidu Wenku can be said to be an ‘industrial revolution’ in content creation, benefiting students, professionals, and creators.


Zhi Xiaobao: From AI Financial Assistant to AI Life Butler


Based on the Ant Bailing model, Zhi Xiaobao not only uses AI to simplify financial management and investment decision-making processes but is also evolving into an AI life butler that can ‘get things done with a single sentence’. From ticket booking to health consultation, Zhi Xiaobao is covering all aspects of users’ lives based on the Ant Bailing model. Especially in the field of financial management, Zhi Xiaobao has more experience in serving financial users, with more professional and rich data analysis, capable of accurately identifying intentions, allowing users to easily obtain precise investment advice and personalized asset allocation plans without deep professional knowledge.


Doubao: A Friend and an Omnipotent Assistant


As an AI assistant launched by ByteDance, Doubao is trying to directly access hardware such as earphones, toys, and smartphones to provide all-weather, all-scenario AI assistant services for users. Doubao, through access to its own Doubao Ola Friend smart earphones, third-party FILL AI earphones, AI toys, and smartphones, is constantly trying and extending the boundaries of AI assistants, while also subtly entering every corner of family, learning, and emotional companionship.


The epoch-making significance of these AI assistants lies in their diverse ways of proving that AI is not only technological progress but also transforming from a ‘high and mighty’ technical tool into a partner close to users, in learning, work, and family life.

Wrapping up, AI, AI, and more AI! The year 2024 is one to remember, not only for the numerous astonishing products it brought forth, but also because these products collectively defined a groundbreaking technological milestone.


The OpenAI o1 model has propelled AI large models from generation to reasoning, revealing new potential in artificial intelligence; the release of HarmonyOS NEXT marked a step towards independence for domestic operating systems on the global stage; and the emergence of Sora is rewriting the possibilities of AI. In the consumer technology sector, OPPO Find X7 began redefining the mobile experience with AI, while Honor MagicOS 9.


0 gave smartphones a prototype of ‘autonomous driving’. The sudden appearance of ‘Black Myth: Wukong’ has also led the world to reevaluate Chinese AAA games. It is also worth mentioning the comprehensive implementation of AI smart assistants. The intelligence of Baidu Wenku, the life services of Zhi Xiao Bao, and the deep penetration of Dou Bao across various hardware are all attempts to naturally integrate AI into our daily lives.


Each epoch-making product is a footnote of 2024; every breakthrough is a preview of the future. It is these products that have transcended the boundaries of imagination, outlining the contours of life for 2024 and beyond.




Leave a Comment

Your email address will not be published. Required fields are marked *