Huawei’s latest paper suggests 10 Intelligent World 2035 directions.

The intelligent world report series was launched by Huawei through an online event on September 16, with the theme “Act Today. Shape Tomorrow.” The Intelligent World 2035: Turning Vision into Action and Global Digitalisation and Intelligence Index 2026 were the two research reports that the company officially released at the event.

The Intelligent World 2035: Turning Vision into Action report, which builds upon the trend landscape outlined in 2024 and the 10 megatrends identified in 2025, introduces ten key directions for building an intelligent world for the first time. It systematically addresses the key challenges in the agentic AI era and provides the industry with an actionable roadmap.

The report emphasises that AI is evolving from a standalone tool to a system that is affecting the fundamental processes of operations and society as it expands from the digital to the physical world. It is expediting transformation across sectors and revising the underlying logic of industry.

In the report, David Wang, the Rotating Chairman of Huawei and Deputy Chairman of the Board, stated, “Agentic AI is a critical variable in this transformation.” In order for AI to transform operational paradigms, the foundations of competition, and industrial dynamics, there are numerous critical obstacles that must be resolved. The direction is evident; however, concrete action is necessary to realise our vision. That is the precise focus of this year’s report, “Intelligent World 2035: Transforming Vision into Action.” This new report, according to Wang, commences with the pivotal shifts that have been brought about by the current AI revolution, analyses the key challenges behind these shifts, and ultimately provides actionable recommendations for an agentic world – the construction of intelligent infrastructure to change operational paradigms and user experience.

Agentic AI is facilitating the transition from an application-centric to an agent-centric digital world, as per the report. In this world, agents continuously perceive their surroundings, autonomously reason and make decisions, dynamically invoke tools, and interact with humans in real time. Massive token generation and high-frequency token flow are integral components of each of these procedures.

According to the report, the global annual token consumption will undergo a 100,000-fold increase by 2035, with agents accounting for over 90% of the traffic. This transition will necessitate significant demands on cloud infrastructure, networking, and computing, which are significantly greater than those of conventional Internet applications. The report delineates ten critical directions in accordance with this observation:

1. Developing AGI with general capabilities and autonomous cognition for any task: The key to AGI is to go physical, which necessitates advancements in scientific intelligence, embodied intelligence, and language intelligence.
2. The industry must transition to SuperPoD computing systems in order to increase token output efficiency, in addition to boosting chip-level computing. This will involve scaling computing clusters by 100 times and reducing agent task costs by 1,000 times.
3. Innovative storage and memory systems that are causally accurate and have reliable traceability to facilitate the transition from data to knowledge and memory.
4. Intelligent connectivity infrastructure that prioritises experience and incorporates AI into communication systems.
5. Tau Scaling Law: Investigating chip design methodologies and tools to co-optimise energy efficiency and latency by replacing geometric scaling with time scaling in order to drive semiconductor evolution.
6. Agent OS: Transitioning from the utilisation of the environment to autonomous evolution and the establishment of an agent-native technical foundation, with its capabilities summarised as: (Coordination × Execution × Memory × Connectivity) ^ Evolution.
7. Establishing a new paradigm for the seamless adaptation and ubiquitous intelligence of devices to facilitate the provision of pervasive intelligent services across devices and scenarios.
8. The ultimate safety and self-learning in intelligent driving to encourage the evolution of vehicles into mobile embodied agents.
9. Overcoming thermal and power constraints to increase the computing capacity of AI data centres.
10. Ensuring the security and privacy of autonomous agents in order to safeguard the intelligent world.

As stated in the report, it is imperative to establish a token-centric, end-to-end intelligent infrastructure that encompasses token generation, delivery, consumption, and security in order to confront these challenges in the agentic AI era. In order to optimise token production efficiency, it is necessary to construct SuperPoDs and open ecosystems, establish intelligent connectivity infrastructure to guarantee seamless token flows, and implement agentic-native cloud platforms to expedite the adoption of AI across various industries. In the end, these practices will transform tokens into perceivable intelligent services, thereby transforming both personal user experiences and enterprise operational paradigms.

In order to conclude the report, Huawei suggests five initiatives: adhere to the principle of tech for good, maintain a long-term perspective, prioritise people, remain dedicated to openness and collaboration, and adopt a systems approach. The objective is to convert technological innovation into industry advancement.

At the occasion, the Global Digitalisation and Intelligence Index 2026, which was jointly developed by Huawei and the Institute of Economics at Tsinghua University, was also released. The index is designed to offer countries a roadmap to an intelligent economy. The report predicts that AI will generate over US$27 trillion in cumulative economic value over the next five years, incorporating insights from experts in economics, sociology, and ICT.

Additionally, the compound annual growth rate (CAGR) of global investment in digital and intelligent infrastructure is anticipated to exceed US$4 trillion by 2030, with a projected growth rate of 19.14%. This will establish a robust foundation for the ongoing development of the intelligent economy.

The report reframes the fundamental variables of the production of the digital and intelligent world as data, ICT talent, and digital and intelligent technologies. It emphasises the importance of integrating networking, computing, and storage to establish a foundation for optimal industrial upgrading. The report evaluates the digital and intelligent advancements of 90 countries that were surveyed and organises their advancements into three stages: Builder, Adopter, and Frontrunner. This information is intended to provide policymakers with a clear path to an intelligent economy. The report also highlights that quantitative assessment enables each nation to identify its own areas of focus and establish a customised development pathway that is tailored to its specific circumstances, despite the absence of a universal paradigm for the intelligent era.