AGI, Agents, Autonomous Driving... Top 10 Technology Predictions for the Next Decade

Recently, Huawei Technologies Co., Ltd. released the report "The Intelligent World 2035", which analyzes how the top ten technological trends in the next decade, including generative artificial intelligence, AI agents, human-machine collaborative programming, multimodal interaction, autonomous driving, and new energy, will profoundly transform various industries. The report suggests that AGI (Artificial General Intelligence) will be the most transformative driving force in the next decade. We are at the cusp of a technological paradigm shift and are gradually moving towards an intelligent world where the physical world and the digital space are integrated in real-time, and the virtual and real coexist.

Source: Report "The Intelligent World 2035"

The following is an excerpt from the report:

Part One: Advancing with AI - The Top Ten Technological Leaps towards the Intelligent World 2035

In the next decade, significant leaps will occur in information technology. This is not a linear development of a single technology but a systematic leap driven by core elements such as data, intelligent technologies, perception and interaction, computing power, networks, and new energy, which deeply permeates both the digital and physical worlds. It is a complex systematic project. The following are the top ten technological trends towards the Intelligent World 2035:

Trend One: Entering the Physical World is the Inevitable Path for AGI Formation

AGI will be the most transformative driving force in the next decade, but it still needs to overcome many core challenges to achieve a breakthrough at the AGI singularity. Entering the physical world is the key path for AGI. Through real-time interaction between physical entities and the environment to achieve the integration of perception, cognition, decision-making, and action, intelligent agents can perceive the world with their "bodies" like humans and grow through interactive learning, thus better adapting to the environment and solving complex tasks. In the future, the focus will be on laying a solid foundation for the realization of AGI from three aspects: the accumulation of multimodal data, the refinement of core capabilities, and the improvement of cognitive principles.

Trend Two: From Execution Tools to Decision-Making Partners, AI Agents Drive the Industrial Revolution

Looking back at history, early agents were information systems focusing on perception and did not have complex decision-making and planning capabilities. In the next decade, agents will develop into action systems focusing on practice.

On the business level, in the next decade, agents will drive a paradigm revolution in various industries. In the early stage, agents will focus on improving operational and office efficiency, acting as customer service, sales, and office assistants, giving rise to an industry worth hundreds of billions of dollars. In the middle stage, they will transform production methods. For example, in fields such as AI drug R & D and professional service consulting, they will significantly reduce R & D and production costs and are expected to create an industry worth trillions of dollars. In the long term, agents will reshape products and experiences. New products such as AI PCs will trigger an office revolution, and embodied robots will become people's personal assistants, expected to open up an industry worth tens of trillions of dollars.

Trend Three: Human-Machine Collaborative Programming Reimagines the Future of Software

With the rise of large models and agents, software development is undergoing a comprehensive reconstruction. It is not only an upgrade of tools but also a systematic change in entry points, roles, processes, architectures, security, quality, ecosystems, platforms, organizations, tools, and collaboration. Ordinary application development will gradually be replaced by agents. Users can generate applications by defining requirements in natural language, forming "malleable software". For professional software development, such as operating systems, financial core systems, and industrial control, it will still be led by professional developers, but the toolchain will be deeply embedded with AI to improve efficiency and verification capabilities.

The essence of the reconstruction of the software development model is "the balance between intelligence and governance". In the next decade, software engineering is not simply "AI replacing humans". Instead, it is a collaborative model between humans and agents: AI undertakes execution and automation, while humans are responsible for design, governance, and value judgment. Agents emphasize both intelligent development and governance.

Source: Report "The Intelligent World 2035"

Trend Four: AI Upgrades in the Mirror World, and New Interactions Open up Immersive Experiences

In the future, with the maturity of multimodal technologies, interactions will quickly move towards three-dimensional space and naturalization. In the short term, micro-gesture interaction technology will become a hot topic in the industry. In the long term, brain-computer interfaces will also be incorporated into the multimodal framework, enabling human-machine collaboration to truly enter the stage of "intuitive interaction".

The update of the interaction paradigm opens the door for the integration of the virtual and real. To further deepen it, the key lies in the comprehensive upgrade of the sensory dimension experience, which will bring more innovative interaction devices, such as rings that can sense delicate gestures, necklaces that can sense pulses, ultra-light 3D glasses, and brain-computer devices. In the future, terminal devices will allow users not only to "see" and "hear" but also to "touch", "smell", and "taste" the digital world, thus achieving a truly full-sensory immersive experience.

Trend Five: The Mobile Internet Ecosystem Shifts from APPs to Multi-Agent Collaboration

In the APP era, the ecosystem is centered around applications. The essence of the ecosystem is "people seeking services". In the multi-agent era, the ecosystem is centered around agents. The essence of the ecosystem is "services seeking people" and even "services executing automatically".

The human-machine interaction interface will move towards the SMUI space multimodal interaction mode. The mobile Internet will transform into an action network. "End-to-end (E2E) task success rate" will become the most core indicator of action ability, directly determining the control of the user entry in the agent era. The traditional advertising model based on consumers' attention is no longer suitable for the multi-agent collaborative ecosystem. More direct value exchange models will emerge, such as intelligence as a service (API and Token calls) and payment based on the success of entrusted tasks (paying according to task success rate rather than duration).

Trend Six: Embodied Intelligence Bridges the Gap and Forms Multiple Trillion-Dollar Industries

In the future, embodied intelligence will be applied in three major industry fields: intelligent driving, intelligent robots, and the low-altitude economy.

Regarding the roadmap for autonomous driving, we expect to start the trial commercialization of Level 4 in late 2027, achieve large-scale application of Level 4 in some scenarios in 2030, and finally reach Level 4+ unmanned driving in most scenarios in 2035. At the same time, we will start the exploration of trial commercialization of Level 5.

Around 2035, with the expansion of mass production, the price of household robots will be less than $10,000. Cost will no longer be the key factor restricting the popularization of robots, and the robot industry will enter an explosive period.

Source: Report "The Intelligent World 2035"

The most critical factor restricting the popularization of the low-altitude economy is the gap in battery energy density. The current battery energy density cannot support several hours of low-altitude flight. In the future, high-energy-density solid-state batteries, multi-source coupling of solar-hydrogen-electric energy, and hybrid power supply of batteries and liquid generators are the key breakthrough directions for the development of the low-altitude economy. We expect that by 2035, the vision of families owning private aircraft may become a reality, and urban transportation will enter the "three-dimensional" era.

Trend Seven: Breaking through the von Neumann Architecture, New Computing Power Meets the Massive Computing Power Demand

Breaking through the singularity of computing power cost will determine the development process of AI. The computing power singularity is when the ratio of the value of AI applications to the computing power cost exceeds 10 times. If the computing power cost singularity can be broken through in a wide range of AI application fields, AI will experience explosive growth.

The traditional von Neumann architecture faces an energy efficiency bottleneck caused by the separation of storage and computing. Against the background of the slowdown of Moore's Law, simply relying on silicon-based semiconductors to increase computing power can no longer meet the exponential growth demand, and it is also difficult to meet the cost requirements. The next decade will be a critical period for the computing field to move from "quantitative change" to "qualitative change": Not only will the computing power density achieve exponential growth, but also through the collaborative innovation of materials, processes, architectures, and paradigms, a new computing power pattern in the "post-Moore era" will be constructed, which will not only solve the bottleneck constraints of traditional technological paths but also provide more efficient and low-carbon computing power support for fields such as the digital economy, artificial intelligence, and scientific research, relieve the global energy supply pressure, and accelerate the progress of human society towards a "smart and low-carbon" future.

Trend Eight: Data is Intelligence, and Agentic AI Drives the Change of Storage Paradigm

With the explosive growth of data access demand for large model training and inference, a large amount of resources once regarded as "cold data" are being reactivated. We predict that by 2035, the proportion of warm data is expected to exceed 70%. The traditional three-layer data structure will gradually evolve into a two-layer structure of "hot-warm - warm-cold", with the ratio approaching 3:7.

In the AI era, the value of data has received unprecedented attention, and there are also challenges in cost and efficiency for the long-term retention of data. The long-term retention of data is not for "hoarding" but for "instant insight". Therefore, the efficient processing of massive "warm data" is the key prerequisite for the large-scale release of AI value.

By 2035, the value of data will no longer come only from its scale or storage form but will be fully released through dynamic activation, scenario-based supply, and semantic storage. At that time, data will not only carry memories but also become the "new fuel" to drive the leap of civilization.

Trend Nine: Jumping from the Mobile Internet to the Agent Internet, Building an Intelligent Interaction Bridge from the Physical Space to the Digital Space

The biggest change in the next decade will be the new connection demand of hundreds of billions of agents. We believe that a next-generation network for agent interconnection is about to emerge. The characteristics of the agent internet are Time-In-Real + hyper-dimensional space (H-D multi-dimensional information). Agents and people are both producers and consumers.

Trend Ten: Token Manages the Energy Network, Making Intelligence the "Nervous System" of Energy

It is foreseeable that in the next 5 - 10 years, global power facilities and energy demand will be the core factors restricting the rapid development of AI. The intelligent era expects a breakthrough in energy technology at the singularity. In the future energy network, Token will become the basic element of energy management. Artificial intelligence technology enables each joule of energy in the future to be defined by Token and endowed with programmable human will, thus realizing the ability of "perception + decision - making + action" for each basic unit in the energy network and completing a revolutionary change in the energy carrier.

With the continuous development of new energy technologies, it is expected that by 2030, the installed capacity of wind and solar new energy will exceed that of traditional fossil energy. By 2035, the power generation of wind and solar will exceed that of traditional fossil energy and become the main power source. It is expected that by 2035, the global installed capacity of thermal energy storage will reach about 500GW, the combined heat and power efficiency of the thermal energy storage system will be increased to 80%, and the power generation from cold and heat storage will become the "stabilizer" of the new power system. Hydrogen energy will have large-scale application opportunities in the fields of heavy trucks, ocean shipping, and industrial fuel substitution, and the global market scale will exceed $1.5 trillion.

Source: Report "The Intelligent World 2035"

It is expected that by 2035, the controllable nuclear fusion experimental system is expected to achieve independent power generation. Once a breakthrough is made in controllable nuclear fusion technology, it will achieve the ultimate reconciliation between AI and energy, and human society will enter a new era.

Click on the cover to subscribe to the annual magazine.

Part Two: Winning with AI, Accelerating the Leap in All Production and Life Scenarios

I. Healthcare: Computing for Health, Preventing Diseases in Advance, and Improving the Quality of Life

Looking forward to 2035, during the health protection period, sensors and wearable devices will enable real-time perception and prediction of individuals' states. AI-driven prediction and intervention means will enable more than 80% of chronic diseases to be effectively prevented, thus effectively extending people's healthy lifespan. During the disease response period, cross-modal AI will integrate medical images, genomics test results, and electronic medical records to provide individualized diagnosis and treatment suggestions for doctors. At the same time, AI-driven drug R & D and virtual clinical trials will significantly shorten the new drug launch cycle and quickly match patients with the optimal drug combination.

In 2035, healthcare will no longer be just about "treating diseases and saving lives" passively. Instead, it will serve as the underlying ability for social operation and lay the foundation for humans to move from "extending lifespan" to "improving the quality of life".

II. Education: Human-Machine Collaborative Teaching Benefits Everyone's Growth

With the development of technology, in 2035, AI technology will be deeply integrated with education, forming a new human-machine collaborative education and teaching model of co-cultivation, co-teaching, and co-learning between humans and machines. Global education will fully enter the era of smart education. In the future, the education system will optimize teaching resources, innovate teaching models, and create a smart teaching environment through ICT technology, especially AI technology, to comprehensively improve educational equity and quality, meet the needs of individual and social development, and enhance the global education level.

By 2035, more than one billion students worldwide will use smart learning assistants for daily learning. At the same time, more than 50 million twin smart teachers will assist in teaching, and more than 80% of smart classrooms will be upgraded to intelligent twin classrooms.

Source: AI-generated

III. Transportation: Experiencing the Mobile "Third Space"

By