Connect 2023: Huawei reveals its plan to help China become a global leader in AI and computing

As part of its effort to become the partner of choice for individual, corporate, and government users throughout the world, Chinese tech giant Huawei has revealed intentions to develop AI-powered solutions across several industry verticals as well as develop a capable computing “backbone.”

Sabrina Meng, the Deputy Chairwoman, Rotating Chairwoman, and CFO of Huawei, reaffirmed the firm’s commitment at a keynote speech and unveiled a whitepaper titled “Accelerating Intelligent Transformation.” Meng spoke about the firm’s ongoing initiatives to delve deeply into fundamental AI technologies and create a strong computing foundation for China.

To create a favourable environment for a thriving ecosystem, we will maintain enhancing the interdependence between hardware, software, chips, edge devices, and the cloud. Our ultimate objective is to assist many businesses in meeting their varying AI computing needs.

Meng said that the business was interested in being “another option for the world” to highlight its commitment to assuming leadership across the globe.

Huawei’s All Intelligence strategy was created to assist all industries in maximizing the new strategic opportunities given by artificial intelligence as it becomes more prominent in the world arena and as its impact on various industries continues to expand.

At this year’s event, Huawei also unveiled a reference architecture for fostering intelligent transformation, along with a number of associated products and solutions. The company’s latest white paper, Accelerating Intelligent Transformation, which provides useful guidance and references to help sectors make the most of intelligence, includes this reference architecture.

Sabrina Meng, Huawei’s Deputy Chairwoman, Rotating Chairwoman, and CFO, speaking at Huawei Connect 2023

Reference to the Huawei Connect event

Annually, Huawei hosts Huawei Connect, a premier event for the world’s ICT sector. It is a forum where the business shares cutting-edge solutions that it has been developing all year.

Huawei used its premier event in 2016 as a stage to showcase its vision for an intelligent world. The business requested help from its global partners in creating a vibrant and diversified tech environment. The company declared its intention to be among the top five cloud service providers in the world the following year, in 2017.

The business unveiled its AI plan at the Connect 2018 conference, outlining how it would become a worldwide power in 5 years. The business then revealed significant developments in its computing strategy at its 2019 Huawei Connect event.

Due to the COVID-19 epidemic, which had a huge impact in China in 2020, this year’s event is the first actual Huawei Connect hosted in Shanghai.

The first virtual Huawei Connect took place in 2020 and was streamed from Shanghai with virtual participation from all across the world. The event’s chairman, Guo Ping, declared that Huawei is generating new values by developing solutions that would draw from the synergies of five different technological areas: 5G, connectivity, cloud, AI, computing, and industry applications.

Revealing a new architecture that supports AI

David Wang, Executive Director of the Board and Chairman of the ICT Infrastructure Managing Board of Huawei announced the release of the new Atlas 900 SuperCluster, the newest member of Huawei’s Ascend range of computing products, in a related keynote presentation. Massive AI foundation models with over a trillion parameters can be trained on the Atlas 900.

The two-layer switching network of the SuperCluster can link up to 2,250 nodes per cluster, or 18,000 NPUs, without oversubscription using high-density 800GE ports.

The new Atlas 900 SuperCluster unveiling by David Wang, Executive Director of the Board and Chairman of the ICT Infrastructure Managing Board

Wang also disclosed that the business had improved system reliability at the component, node, cluster, and service levels by systematically leveraging its strengths in computing, storage, network, and energy.

It’s crucial to have a reliable system while training large foundation models. The cluster’s capacity to enable continuous model training has now been effectively increased from a few days to a month or more, he noted.

“We now stand at the threshold of a new and intelligent world, with vast opportunities and challenges ahead of us. We need to work together, dig deep into industry-specific scenarios, and build a solid computing backbone to power countless new AI models and applications. Together, we can help all industries go intelligent, and help them do it faster”,

David Wang

Wang oversaw the launch of nine new intelligent industry solutions from other partners, all of which were based on Huawei’s intelligent transformation reference architecture. These are made to specifically address the requirements of many sectors, including manufacturing, electric power, railways, government, and finance.

Showcasing the Pangu AI model for several applications

The Pangu AI model from the company will be unveiled at Huawei Connect 2023, which is sure to be spectacular. Pangu, which was introduced by Zhang Ping’an, Executive Director and CEO of Huawei Cloud Computing Technologies, focuses on the most urgent requirements that various clients have.

Zhang-Pingan-Executive-Director-of-Huawei-and-CEO-of-Huawei-Cloud

The Pangu Models 3.0 was originally released by Huawei in July. A variety of models and applications can be supported by the company’s 5+N+X decoupling architecture, which was later announced.

Pangu models are now being used in specialized applications for mining, government, automobiles, weather, medical, virtual humans, and R&D thanks to Huawei Cloud’s most recent announcement. The Pangu Vehicle Model, for instance, offers a solution to the challenge of teaching autonomous driving systems to handle a variety of challenging situations. This problem is addressed by the Pangu vehicle big model, which generates automatic annotations for simulated circumstances like dusty mining settings, steep inclines, and fast bends.

The training period for autonomous driving corner cases has been reduced as a result from over two weeks to just two days.

“We now offer another option for global customers, partners, and developers to build, train, and use AI models,”

Zhang Ping’an

The CodeArts Snap, a sophisticated programming assistance for developers, is another intriguing aspect of the Pangu approach. It is comparable to the GitHub co-pilot, an AI-driven coding chatbot owned by Microsoft.

CodeArts Snap can generate code through a single dialogue, suggest code, automatically comment out and generate test cases with a single click, and launch services/apps with a single instruction. It has been trained with 76 billion lines of high-quality code and 13 million technical documents.

Huawei Cloud Ascend AI cloud service and more

Additionally, he declared the opening of the Huawei Cloud Ascend AI cloud service. He claims that starting today, it is accessible for purchase and use on the Huawei Cloud website. Additionally, he declared that the Huawei Cloud website now has a separate zone for the Ascend AI cloud service’s core models. Open-source models like LLaMA and GLM are now accessible to enterprise users and developers.

Cloud services from Ascend AI employ well-known AI frameworks like PyTorch and TensorFlow. Using the given end-to-end migration tools, 90% of operators in these frameworks may transition to Ascend without any issues.

In a similar vein, Huawei Cloud CTO Bruno Zhang stated that the company has started a lengthy research project that will lead to a dynamic innovation of the current software and hardware structures to specifically support the deployment of AI as it works to become the cloud of choice for the AI era.

The result, in his opinion, is the development of a model that satisfies user requirements in terms of system design, data processing, model training, application development, and layers of cloud-based AI services.

For instance, Ascend AI cloud service can integrate with existing data centers utilizing Huawei’s Cloud Stack Stack 8.3, a hybrid cloud platform, if an organization needs a specialized adaptable AI platform. According to Bruno, the general public can now use this service.

One can only conclude, given the company’s commitment to “tackle big challenges head-on” as stated by Zhang Ping, that Pangu models, when included in Huawei Cloud products and services, significantly increase productivity across industries.

Author

Samson Ayodeji

I am literally obsessed with technology. I love writing about the latest news in technology in different markets, segments and sectors across the world.

View all posts