FREMONT, Calif., June 14, 2018 /PRNewswire/ -- 2018 Baidu Data Center Partner Conference was held in Beijing on June 7. Inspur, a strategic supplier partner of Baidu, was awarded the annual "Technology Innovation Award" at the event. This is the third time that Inspur has won this award.
Inspur is one of Baidu's major server suppliers and key partner in infrastructure innovation. Inspur and Baidu started their collaboration 7 years ago, and hundreds of thousands of Inspur servers have since been installed at Baidu Data Center. The SR-AI Rack and ABC all-in-one machine jointly developed by Inspur and Baidu for large-scale offline training have been widely applied in search engines, AI, image recognition and speech recognition, providing fundamental computing support for Baidu's business innovation.
JDM drives partnerships from simple supply-and-demand to deep collaboration
IDC data shows that Inspur retains the leading share in the Internet industry for 4 consecutive years and Baidu's purchase has always been a major contribution to Inspur's Internet share. Currently, with its JDM (joint design manufacturing) business model, Inspur has been widely recognized by clients in China and North America. Unlike traditional OEM or ODM, the JDM model is based on the infusion of Inspur and client's industry chains, offering customized services to clients covering the whole service chain from R&D, manufacturing, supply, implementation to operation and maintenance.
It was through the alliance with Baidu that Inspur gradually developed the JDM model to maturity. In 2012, Inspur launched rack-scale server SR3.0 and took this opportunity to take root in Baidu's business. Since then, Inspur and Baidu have continued their partnership till present and deepened and expanded their collaboration from purchase and product customization in the past to business chain fusion today.
The clearest indication of the deepening alliance between Inspur and Baidu is that Inspur's product customization service to Baidu has penetrated into the disruptive innovation of system architecture. In 2017, Inspur and Baidu jointly developed the SR-AI rack-scale server for large-scale offline training. As the first AI solution based on PCIe Fabric technology, this SR-AI Rack can effectively improve the efficiency of AI offline training as each node supports a maximum of 16 GPUs, and each system supports up to 64 GPUs. This AI solution has achieved satisfactory application results in Baidu's unmanned driving and image search businesses.
At the same time, the collaboration between the two parties also advanced from product technology innovation to back-end supply and quality. As early as a few years ago, Inspur and Baidu brought about supply chain platform docking, achieving online transmission of material kitting – order, production schedule, arrival schedule, order status, assets, and logistics information. Thanks to the flexible production capacity of Inspur's newly launched smart plant and the joint optimization of the two parties, the delivery efficiency of Baidu server has more than doubled.
Collaborative innovation promotes AI commercialization
In September 2017, Inspur and Baidu released a series of ABC all-in-one machines. This series combines the cluster management software independently developed by Baidu and the AI computing hardware platform of Inspur. This out-of-box and all-in-one solution supports all deep learning frameworks, and the embedded mature algorithm model and cloud management technology.
Prior to this, the cooperation between the two parties has been feature-driven development and supply of IT infrastructure. This time, the two parties have further expanded their cooperation to the marketing field. On one hand, Baidu has leading the AI technology, such as image recognition and voice recognition. Among 10,000 steel pictures recognized by the trained Baidu steel quality recognition program, only 3 pictures are misidentified, which far exceeds artificial efficiency. One the other hand, Inspur has global marketing and user coverage capabilities. The alliance of Baidu and Inspur can expedite the adoption of mature AI solutions into traditional industries, promoting large-scale AI commercialization.
Building an ecological environment to promote industrial upgrade
Though the businesses of CSPs may vary, they all have large IT infrastructures, for which they share a common core competence - data center business. This business not only provides computation basis for various enterprises, but can also provide external servers to achieve resource marketability.
To retain a long-term competitive edge, CSPs should not only focus on cost or technology, but also develop a competitive industrial ecosystem. This is the primary cause for the rapid development of open source hardware communities such as OCP, ODCC, and Open19. Under this circumstance, what CSPs need is not a supplier but a partner who can provide greater value. It is also for this reason that Inspur's JDM model has been highly recognized. In the future, Inspur hopes to further deepen the integration with the user's business chain and play a bigger role in the user's data center ecosystem. Moreover, Inspur wishes to join hands with leading Internet service providers such as Baidu to further promote the transformation and upgrade of the "Internet+" industry.