Tencent Cloud Launches Full-Stack AIGC Solution to Accelerate AI Adoption Across Industries
-
On September 8, at the 2023 Tencent Global Digital Ecosystem Conference's AIGC Application Session, Tencent Cloud officially launched its full-stack AIGC solution, offering enterprises trustworthy, reliable, and accessible end-to-end AIGC content services. This initiative aims to assist businesses in exploring the path to AGI and seizing new opportunities in the AI 2.0 era. The company also collaborated with industry partners to share practical AI application scenarios and explore new pathways for deep integration of AIGC across various sectors.
Xu Huabin, Vice President of Tencent Cloud, stated in his opening remarks: "Currently, AIGC has been widely adopted across industries. The key focus now and in the future is how to effectively implement this technology in business operations to achieve genuine improvements in productivity and efficiency. Tencent Cloud is committed to partnering with enterprises as a technical assistant to develop high-quality AIGC applications and deliver smarter, more convenient services to users."
Tencent Cloud's Full-Stack AIGC Solution Ensures 'Trustworthy, Reliable, and Accessible' Services
Mao Dehui, Tencent Cloud's industry solutions expert, noted that as the AI 2.0 era accelerates, businesses face numerous challenges in implementation, including poor data quality, insufficient large-scale computing power, high inference costs, and compliance requirements.
To help industries accelerate digital and intelligent transformation and address AI adoption challenges, Tencent Cloud introduced its full-stack AIGC solution. Leveraging Tencent Cloud's AI capabilities in large models and machine learning, the solution upgrades its one-stop MaaS (Model-as-a-Service) offering and establishes a comprehensive AIGC content security framework. It provides end-to-end development services—from data annotation, training/fine-tuning, evaluation, and testing to deployment—ensuring AIGC services are "trustworthy, reliable, and accessible."
Computing Power: Tencent Cloud's next-generation High-Performance Computing Cluster (HCC) delivers high-performance, high-bandwidth, and low-latency intelligent computing support for large model training across three technical dimensions: computing, storage, and networking. Equipped with the latest-generation computing cards and leveraging the 6U ultra-high-density design of its self-developed StarLake servers, it ensures high-performance computing. High-throughput storage is achieved through CFS Turbo and COS (Cloud Object Storage), while the self-developed 3.2T RDMA Starlight network guarantees high-performance, lossless networking. The cluster also features an observability system for real-time GPU and network monitoring with automatic fault recovery, ensuring high stability.
Additionally, Tencent Cloud systematically improves resource utilization through hybrid deployment strategies, including GPU & CPU co-scheduling, GPU online/offline co-scheduling, and local disk/distributed cache co-scheduling. The HCC cluster lays the computational foundation for AIGC adoption across industries, making AIGC services more reliable.
Compliance: Tencent Cloud offers dataset preprocessing, automated review services, and customized recognition solutions. By training and fine-tuning with AIGC large models, it builds a robust content moderation platform to intelligently filter sensitive content, ensuring end-to-end compliance. This effectively addresses content compliance challenges in AIGC implementation, making services more trustworthy.
Training & Inference Efficiency: Tencent Cloud provides end-to-end acceleration capabilities, including dataset download acceleration, training/model data read acceleration, checkpoint write acceleration, distributed training acceleration, and inference acceleration. These systematically resolve efficiency issues, reducing costs through improved performance. For example, its semi-managed MSP tool and COS enable cross-region replication, facilitating low-cost transfers of overseas datasets to domestic servers.
Mao Dehui added, "As model complexity increases, inference latency rises significantly. Meeting performance demands requires exceptionally high GPU resources, driving up per-QPS inference costs. Currently, large models are primarily used for productivity due to high service costs, limiting them to high-value scenarios. If costs could drop to 1/10 or lower, large models could expand from productivity to entertainment, content, and all user interfaces." Tencent Cloud's full-stack acceleration capabilities help enterprises improve efficiency and reduce costs on the path to AGI, making AIGC services more accessible.
Taking the large model startup MiniMax as an example, its Vice President Wei Wei introduced that through Tencent Cloud's new-generation high-performance computing cluster (HCC), MiniMax completed a technological infrastructure upgrade, reducing overall cloud costs by at least 20%. Tencent Cloud's firewall, WAF, and other security products ensure stable business operations and data security. Currently, the MiniMax open platform has onboarded over 100 paying customers, covering more than ten industry scenarios such as office collaboration, interactive entertainment, customer service, search, and education.
As a leading Chinese internet platform expanding globally, Kunlun Wanwei shared insights on its newly launched Tian Gong AI search, which integrates large models to enhance user search experiences through humanized and intelligent approaches. Lin Jian, Deputy General Manager of Commercialization at Kunlun Wanwei, explained that Tian Gong AI search allows interaction in any natural language, understands users' true search intent, provides effective answers, and supports over 20 rounds of follow-up interactions. It will soon feature multimodal search capabilities, including images and voice. Kunlun utilizes Tencent Cloud's elastic computing power, databases, and big data services for AI search and AI music composition, enabling rapid iterations in AI search capabilities through multi-generational model training and optimization, as well as data collection, cleaning, and organization.
At the event, Li Xiang, Deputy Director of Tencent Cloud's Industry Architecture, shared how AI-native capabilities drive innovation for SaaS enterprises. The emergence of AIGC has accelerated the shift from process-driven to intelligent SaaS development. Tencent Cloud's AI-native capabilities—such as image generation, video generation, speech-to-text, and content compliance—provide technical support for internet enterprises in scenarios like content creation, intelligent Q&A, recommendation systems, and smart summarization.
Taking CRM as an example, Wang Chaohui, Vice President of Products at Sales易, noted that large models have ushered in a new era for intelligent customer service robots. These robots can not only identify customer intent but also analyze context and engage in multi-turn conversations, guiding users to solve problems step-by-step. This marks a departure from traditional menu-based interactions, offering users an immersive and interactive experience.
In the PaaS domain, Xu Hao, Vice President of Kingdee China and General Manager of the Cangqiong Platform Solutions Division, explained that through collaboration with Tencent TI-ONE, Cangqiong GPT provides enterprises with a new technological foundation combining cloud-native and large-model capabilities, reshaping digital competencies. Cangqiong GPT has become an AI assistant that understands enterprises and a large-model platform tailored for management. Kingdee Cloud·Cangqiong has ranked first in China's market share for three consecutive years and is the only Chinese product in the global top 10 for high-productivity aPaaS.
ZhenDao Group (Marketingforce.com), a global leader in SaaS intelligent marketing cloud platforms, has long dominated China's marketing and sales SaaS solutions market. Through continuous exploration and innovation in AIGC, ZhenDao has driven comprehensive digital transformation for partner enterprises. According to CTO Xu Youfu, ZhenDao integrates AIGC into high-value marketing and sales scenarios, creating an AI-driven value loop that exponentially enhances marketing effectiveness. ZhenDao's solutions include "AIGC + Smart Audience Targeting" and "AIGC + Smart Marketing Canvas Strategy Generation," leveraging big data for intelligent implementation. Additionally, "AI + Digital Human" technology addresses marketing challenges in short videos and live streaming, helping enterprises create personalized digital IPs free from human limitations.
AIGC Brings Development Opportunities, Tencent Cloud's AIGC Solutions Accelerate AI Adoption Across Industries
While AIGC accelerates intelligence, it also creates new opportunities for innovation. Zhu Lihua, Chief Technology Officer of Momo Group, shared how AIGC is applied in Momo's open social scenarios. Addressing users' desire for interaction while avoiding real photo uploads, AIGC generates refined virtual avatars and stylized dynamic images based on users' real appearances, striking a balance between "resemblance" and "privacy." This explores an innovative path combining "image generation + open social networking."
Li Zhifei, Founder and CEO of Mobvoi, introduced the company's multimodal general large model "Sequence Monkey" and the one-stop AIGC SaaS services built on it. Li noted that large models have entered a phase focused on applications and implementation. Commercially, AIGC companies can adopt a unique business model—serving SMBs or Professional Consumers (ToSMB/ToPC)—distinct from traditional ToB or ToC approaches. "Sequence Monkey" is a multimodal, learnable Agent-based large model. Through vertical integration of AIGC applications and large models, Mobvoi has created a one-stop AIGC product matrix for content creators, pioneering this new business model.
According to Ren Yongliang, founder and CEO of China's leading online mental health community CeCe, the country faces a severe shortage of professionals in the mental health sector, and the quality of services needs significant improvement. AIGC-based mental health services have effectively filled this gap. The core of mental health services lies in language and dialogue technologies, which naturally align with the principles of AIGC. Leveraging this, CeCe has introduced virtual experts, 3D psychological sandbox therapy, and XiaoCe AI for end-users, enabling them to access emotional support anytime, anywhere. Currently, millions of users on the CeCe platform have utilized AIGC services and provided positive feedback.
Building on the widespread adoption of AIGC across industries, Tencent Cloud has officially launched its comprehensive AIGC solution. This solution leverages four key generative capabilities—image generation, video generation, text generation, and speech generation—to offer over 10 standardized applications, including AI painting, portrait transformation, virtual idol creation, and AI music composition.
The solution covers content creation, editing, and operational scenarios across multiple internet industries, effectively assisting businesses in driving innovation. Liu Zhenjiang, Deputy Director of Tencent Cloud's Industry Architecture, noted that the AIGC solution has helped clients like CeCe, Kuaikan, Shanmeng, Jingling, and Light Novel implement AIGC applications.
In the tourism sector, Chen Hui, Executive Vice President of Mafengwo, shared that the team is using Tencent Cloud's large-scale models to integrate Mafengwo's content, product, and Q&A databases into a tourism knowledge base. Further data processing will fuse this into a tourism knowledge engine, creating a "travel guide + AI" smart travel companion to provide reliable and personalized services for travelers.
For the entertainment industry, Dr. Wu Bin, Senior Director of Tencent Music's Tianqin Lab, discussed the application and development of AIGC in entertainment. Through products like large language models, visual AIGC, audio AIGC, and large model acceleration, Tencent Cloud has significantly enhanced users' audiovisual experiences.
In the design industry, Wang Baochen, founder and CEO of Chuangkit, highlighted how AI has transformed content production and marketing. As one of the first domestic AI-powered design tools, Chuangkit AI integrates years of industry knowledge, resource data, and cutting-edge AIGC technology to become a more user-centric design assistant.
AIGC presents abundant entrepreneurial opportunities in entertainment. Liang Jian, CEO of RightBrain Tech, shared insights into their product Vega AI's advancements in video creation. He noted that traditional visual content creation had high barriers and supply shortages, but AIGC has revolutionized production efficiency. RightBrain Tech recently secured tens of millions in angel funding in June.
At the event, Tencent Cloud also signed AIGC strategic cooperation agreements with Zhipu AI, ZhenDao, Mobvoi, FaceWall Intelligence, and APUS to explore more AIGC applications and accelerate the advancement of the AI industry.
Miao Junwei, Vice President of FaceWall Intelligence, expressed excitement about the future collaboration with Tencent Cloud, stating, "Our CPM large model has already been implemented in finance, education, and industrial sectors. We remain committed to 'bringing large models to millions of households,' refining standardized processes and tools to efficiently meet customer needs and lower the barriers to AIGC adoption, democratizing its benefits."
Under the strategic partnership, Tencent Cloud and Zhipu AI will collaborate deeply on computing power, focusing on high-performance large model training clusters and improving inference efficiency. They will also combine their strengths to develop diverse cooperation models and offer "Model as a Service (MaaS)" solutions to industry clients, fostering a robust large model ecosystem.
As AI 2.0 gains momentum, AIGC applications have become essential across industries. Moving forward, Tencent Cloud will continue to optimize its AIGC services, helping more industries and businesses adopt large model-based AI technologies cost-effectively. This will enable partners to swiftly embrace AI, drive innovation, achieve business growth through AIGC, and seize new opportunities in the AI 2.0 era.