PPIO Launches New Compute Cloud Product, Driving 1,000x Cost Reduction for LLM Inference | BlueRun Ventures Headlines

On September 4, 2024, "Power to Pioneers: 2024 Distributed Cloud Computing Forum & PPIO AI Product Launch", hosted by PPIO, was held in Shanghai. The forum focused on technological innovation in distributed cloud computing, industry implementation, market trends, and its driving force for the large-scale application of AIGC. During the event, PPIO launched new Serverless and Model API service products, further completing its Paio Compute Cloud product matrix. This marks PPIO's expansion from edge cloud to AI infrastructure, bringing AIGC innovators and developers more cost-effective, elastic, and user-friendly one-stop AI infrastructure products and services.

Yuan Tao, Party Secretary and Chairman of Shanghai Zhangjiang Group, and Hai Jin, Professor at Huazhong University of Science and Technology, Changjiang Scholar Distinguished Professor, and Vice Chairman of the China Computer Federation, attended the forum and delivered speeches. Enqing Xu, Director of the Government and Enterprise Digital Transformation Department at CAICT's Cloud Computing and Big Data Research Institute, and Haibo Chen, Distinguished Professor at Shanghai Jiao Tong University, Deputy Director of the China Computer Federation's System Software Committee, and Chair of the OpenHarmony Technical Steering Committee, gave inspiring presentations that contributed abundant ideas to the development of the distributed cloud industry.

Xin Yao, Co-founder, Chairman, and CEO of PPIO, pointed out in his keynote speech: "Following the fundamental principles of scaling laws, computing power will determine the ceiling of large model innovation and application. In a market landscape where computing resources are costly yet demand is exploding, distributed computing power will become a crucial driving force for the next round of technological revolution, gradually transforming how global computing resources are supplied and utilized. Our core philosophy is to leverage the advantages of distributed cloud to achieve a thousandfold cost reduction within three years, promoting AI accessibility through computing power accessibility, and helping pioneers of the new era create entirely new development opportunities."

Compared to model training, AI inference is more closely tied to industrial applications. After training, large models need to be deployed through inference in real-world scenarios, which directly impacts their usability and practicality. However, the primary challenge facing AI inference currently is high cost — especially as model scales increase, computing demands and costs rise accordingly. Additionally, inference efficiency is generally low, and complex business scenarios and application chains further drive up inference costs.

To address the cost challenges of AI inference and achieve accessible computing power for large model applications, PPIO has developed a Serverless product specifically designed for AI inference scenarios through coordinated innovation across algorithms, systems, and hardware. It provides model deployment, runtime optimization, elastic scaling, and API services, helping customers achieve highly efficient elasticity and optimal runtime costs when customizing models. The launch of the Serverless product allows developers to no longer manage and operate compute servers, instead focusing their energy on core business application logic development and innovation, thereby helping enterprises reduce costs and increase efficiency.

Paio Compute Cloud's Serverless product offers industry-leading automatic elastic trigger latency and single-instance cold start speed

Furthermore, PPIO launched its new Model API service product, supporting full-modality APIs required for AIGC application development. Developers can quickly deploy and invoke large model capabilities without needing expertise in machine learning. Combined with PPIO's powerful distributed computing capabilities and the elastic support of Paio Compute Cloud's Serverless product, the Model API service not only significantly improves AI application development efficiency but also effectively reduces enterprise development and operating costs through a pay-as-you-go model.

PPIO Co-founder and CTO Wenyu Wang announces Model API service pricing

Paio Compute Cloud products can be widely applied across multiple scenarios, providing flexible and efficient solutions. For example, in China Tower's intelligent computing power scheduling and management project, to achieve computing power入园 (park access) management services across widely distributed industrial parks, office buildings, residential areas, entertainment venues, and universities in cities, PPIO built a distributed computing power network enabling elastic computing power scheduling and large model inference acceleration across different regions, times, and businesses. This provides high cost-performance intelligent computing services for AI applications within parks, meeting needs for intelligent data processing, large model deployment, AIGC generation, AI algorithm acceleration, rendering acceleration, and a series of other application requirements.

At the conference, PPIO announced the appointment of Hai Jin, Professor at Huazhong University of Science and Technology, Changjiang Scholar Distinguished Professor, and Vice Chairman of the China Computer Federation, as Chairman of the company's Technical Committee. As a renowned scholar in China's distributed computing field, Professor Jin will provide comprehensive and in-depth guidance for PPIO's distributed technology development strategy and computing power network planning, laying a solid foundation for the company's future research and innovation in distributed cloud and AI infrastructure.

Professor Jin stated: "PPIO's frontier exploration and innovative practices in the distributed cloud field not only demonstrate the profound impact of technology-driven industry transformation but also herald the arrival of an entirely new computing era. I look forward to working with the PPIO technical team to further expand the technological boundaries of distributed cloud computing, promote its deep application across more industries, and help PPIO achieve globally leading distributed computing resource optimization and industry digital transformation."

Meanwhile, the China Academy of Information and Communications Technology, together with PPIO and multiple industry-leading enterprises and research institutions, jointly launched research on the White Paper on Distributed Computing Power Development, aiming to advance distributed computing power technology and inject new momentum into the global digital economy. The white paper will systematically elaborate on the development background, core technologies, application scenarios, market trends, and future directions of distributed computing power. It will comprehensively summarize the practical achievements and technological breakthroughs of distributed computing power globally in recent years, providing clear direction for distributed cloud industry practitioners, technology developers, and policymakers, and promoting the further maturation and implementation of distributed computing power technology.

Beyond this, to drive continuous innovation in AI applications, PPIO has partnered with PingCap, Zilliz, AscentStream, and CSDN to launch the "Pioneers AI Startup Acceleration Program." Targeting AI startups and developers, the program provides comprehensive resources and technical services including distributed computing power, distributed relational databases, distributed vector databases, and distributed message queues, helping startup teams achieve rapid growth on distributed cloud and accelerate the implementation of their innovative outcomes.

Xin Yao, Co-founder, Chairman, and CEO of PPIO, stated: "As a second-time entrepreneur who has walked the entrepreneurial path for 20 years, I deeply understand that the road of innovation is not always smooth. We stand at the threshold of a new era, witnessing wave after wave of pioneers full of dreams — just like that young Xin Yao back then. What they need is not just funding and technology, but trust and encouragement from fellow travelers and those who came before, just as my teacher Professor Jin Hai pointed the way forward for me back then. Salute to the pioneers of the new era — let us together seize innovation opportunities in this new AI era."

PPIO will continue to leverage the advantages of deep industry-university-research integration and ecosystem co-construction, building innovative product services based on distributed cloud, continuously reducing technology infrastructure costs, enabling more AI enterprises and developers to easily afford large model inference expenses, and helping enterprises gain market advantages in AI application innovation.

ZeroG Aircraft Industry Completes Nearly RMB 100 Million Series A Financing, Continuing to Drive Development and Application of New-Energy Aircraft | BlueRun Ventures

Packed House! How Embodied Intelligence Technology Is Reaching Industrial Implementation — Insights from the World Robot Conference BlueRun Ventures Forum

---

Dier Boge Secures Tens of Millions of RMB in Pre-A Round, Focused on Intelligent Robot Workstations for Heavy Industry | BlueRun Ventures

---