Huawei Analyst Summit 2024 Takeaways

During the Huawei Analyst Summit (HAS 2024) hosted in Shenzhen, we had the opportunity to learn about and assess the company’s All Intelligence strategy. Artificial intelligence (AI) has certainly become the hottest technology as of late, and even a mainstream topic admittedly since OpenAI released ChatGPT at the end of 2022. We were not disappointed by the breadth and depth of the AI-related sessions during the event or the scope of the company’s All Intelligence strategy.

Our key takeaways

These are our key takeaways from the myriad of sessions we had the opportunity to attend, from keynotes and breakout sessions, to roundtables, group meetings and one on ones:

• Ambition in AI scope: In our view, it is important to understand the difference between the adoption of AI and generative AI technologies to improve operations and the business opportunities form assisting other organizations in its adoption. Huawei seems to be stating a broad ambition to play in both fronts: i) on the one hand, using AI to automate network operations or in the software running in its products, such as radio access network equipment, and ii) on the other, developing large language models (LLM) like the Pangu models and the Celia assistant, to capture the broader opportunity from companies across many sectors adopting AI.
• Intelligent autonomous networks: China is making significant progress in implementing intelligent autonomous network concepts in operations and management, including maintenance and monitoring, optimization and configuration, and quality assurance among others. We heard from officials from the China Academy of Information and Communications Technology (CAICT) as well as executives from several local carriers, such as China Mobile and China Unicom, who are using Huawei’s Autonomous Driving Network (ADN) offering to achieve level 3 and 4 of autonomy. While admittedly other countries may not be so far ahead in implementing intelligent autonomous networks, this product capability and implementation experience may become a competitive advantage when the time comes.
• Common AI research progress: We were encouraged to see that AI researchers in organizations in China, the US, and Europe seem to be collaborating and learning from each other. Areas such as multi-modality, AI agents, super-alignment and the need for AI paradigms beyond large language models, seem to be shared across borders, so what we heard from Huawei’s experts this week was very consistent with what we hear in other regions.
• Sustainability: Huawei has long pursued a sustainability agenda of “green energy” and “green data centers,” but this is becoming more important with the recent resurgence of generative AI, which is extremely power hungry, and Huawei appears well positioned to handle these conflicting trends.

Aligning to AI adoption trends

Within the Thematics Intelligence team at GlobalData we cover the AI theme in detail, with a multi-sector perspective. Specifically, on generative AI, we envision a technology roadmap with four phases, the transitions of which are driven by milestones such as the resolution of the current accuracy issues of current large language models (i.e., hallucinations) and the achievement of artificial general intelligence (AGI).

Source: GlobalData

As each specific AI use case would require different levels of accuracy and/or an internal capability to hold a representation of the world, we expect AI adoption to be heterogeneous across business functions and industries. Indeed, while we expect the global generative AI market opportunity to grow 80% annually until 2027 to reach US$33 billion, we believe adoption will be uneven across verticals, with information technology, retail banking, government, manufacturing, retail, healthcare, and construction leading the way.

Source: GlobalData

Our analysis indicates that the majority of industrial sectors approach adoption at three levels, which exhibit increasing degrees of complexity.

• Business process digitalization: arguably the low hanging fruit lays in using AI to further automate and digitalize business processes, which can lead to cost savings, increased productivity, or both. From using computer vision to manage inventory in a warehouse, to having a chatbot deal with customer queries in a call center.
• Product and services innovation: a second level of AI enablement can be achieved by innovating in the core products or services of a business, which can lead to a competitive advantage, either in productivity, pricing, or new offerings. Consider the examples of using AI for drug discovery in the pharma industry, or to optimize the air interface (e.g., channel state information (CSI) feedback enhancement, beam management, positioning accuracy) of radio access network equipment in the telecom equipment industry.
• Policy level: a third level that is still very much conceptual at this stage, consists of using AI to extract insights and drive policy at the overall market level. For instance, with the right data privacy protections in place, AI could be used by national or multi-national healthcare organizations to improve outcomes by optimizing the allocation of resources across the “continuum of care”, from primary care to acute care and community social care. Another example would be to enhance treatment protocols based on insights extracted from data at the country level. Likewise, education departments should be able to improve national curriculums and attainment levels by extracting insights from available data, or leveraging AI to tailor the educational experience to each individual student.

We believe most sectors are currently in the first level, often exploring the second, yet this is bound to evolve rapidly as the capabilities of our AI systems continue to improve.

AI enabled Huawei’s products and solutions

As noted earlier, it became apparent during the various sessions we attended that Huawei has an ambition to facilitate at least the first two levels, and possibly the third if implemented, given its close collaboration with standard bodies and regulators. For instance:

• The company has been an early adopter in using AI/ML to enhance its connectivity products. Its 5G New Radio (NR) equipment leverages machine learning for beam management to improve accuracy, and reduce overheads and latency. Beam management is an important part of 5G technology, as by directing radio transmissions using sharp and focused beams, it improves signal strength and reduces interferences between users. As such, any improvement in beam management translates into a better user experience.
• Huawei is a strong supporter of highly autonomous networking technology, which it calls autonomous driving network (ADN). CAICT has now launched the Autonomous Network Pilot Program 2.0, and several leading domestic carriers are involved in the initiative, including China Mobile and China Unicom. At the conference we heard from various operators how using the Huawei ADN Net Master and ICN Master applications, which are powered by the company’s Telecom foundation model and Pangu foundation model, they are able to achieve L3 and even L4 autonomy in network operations and management.

Image source: Huawei / Huawei Cloud Core Network ICN Master screenshot

Image Souce: Huawei / Huawei ADN Net Master

• That said, Huawei’s foray into AI with its Pangu large language models is not limited to its traditional core expertise in the communications industry. Various Pangu models have been trained for very varied scenarios, from forecasting the weather to industry specific knowledge in finance, mining, manufacturing, or telecommunications. As such, it is to be expected that not only will Huawei use them to enhance its own product and service offerings, but that its clients will be able to use them to implement business process digitalization initiatives and even product innovation.

Developments in generative AI are occurring very fast, not only in Shenzhen but worldwide, and competition will certainly not stand still. However, we left the event with the sense that the company has an ambitious and well thought through “All intelligent” product roadmap, and it should find significant business opportunities in many sectors, from industrial verticals such as telecommunications, mining, energy and manufacturing, to less obvious ones such as life sciences and education.

About the author

Josep Bori is research director for GlobalData’s Thematic Intelligence team. In this role, he manages the production and publication of thematic content across the company, with a particular focus on technology research. Prior to joining GlobalData, he was head of TMT content at expert network firm GLG.Before that he worked over 15 years as sell-side equity research analyst covering global technology stocks at firms including Deutsche Bank, Exane BNP Paribas, Atlantic Equities and Berenberg Bank.