2040 - A Foresight

– Kai-Fu Lee, AI Expert, Chairman & CEO of Sinovation Ventures, Author of 'AI Superpowers' and 'AI 2041'

By 2040, we will be living in a new world where cloud computing will become the Modern Stone Age. As processors become smaller and more efficient through neuromorphic and quantum architectures, intelligence will operate locally. Devices such as phones, cars, lenses, and implants will process information on their own. The cloud will function mainly as a coordination layer rather than the core of computation.

Connectivity will return to communication. Humans will interact with machines, and machines will interact with one another. Information exchange will occur as dialogue instead of command. Neural interfaces will connect human cognition directly with digital systems. What began as clinical brain–computer interfaces will become everyday extensions of memory, intention, and perception.

Hardware itself will change. Developments in nanomaterials, memristors, and phase-change transistors will make physical circuits reconfigurable. Devices will be able to modify their architecture through updates. A processor could expand its capacity by activating dormant transistors or redirecting electrical paths. Hardware upgrades could happen through a software patch.

These changes will create new business models. Companies will shift from selling devices to licensing adaptive capabilities. A manufacturer might charge for unlocking additional processing layers, neural bandwidth, or sensor precision over time. Subscription tiers could determine how much local intelligence a user can activate or how personal an AI system becomes. Traditional data centers will evolve into marketplaces for distributed learning, where devices exchange insights without sharing raw data. Value will come from the quality of local computation, not the scale of centralized storage.

Local intelligence could become common by 2030, neural interfaces practical by the 2040s, and adaptive hardware widespread by mid-century. When that occurs, computation will exist everywhere; within objects, systems, and human networks. The distinction between being online and being present will disappear.

– Ginni Rometty, Former CEO of IBM and author of Good Power

Sounds too far-fetched? Think again. 

On-device AI chips already run large models locally on phones and PCs (Apple Neural Engine, Google TPU Edge, Qualcomm Hexagon NPU). Neuralink received FDA clearance for human trials and reported its first implant in 2024. Synchron has transmitted brain signals using an endovascular stent. Meta’s EMG wristband decodes intent from muscle activity. IBM’s analog in-memory AI hardware and HP Labs’ memristor arrays demonstrate learning circuits. Peer-reviewed work on programmable metasurfaces shows antennas and optics that can be reprogrammed in software. These systems exist today in products, clinical studies, or lab prototypes.

One last thing. Digital Continuity business will come into existence. Our behavioral and functional data — collected through productivity tools, communication platforms, and social media — will provide enough information to create digital versions of us after death. These models will replicate a our tone, reasoning style, and decision patterns. Companies already use generative AI to train personal assistants and simulation agents on individual data. The same methods would preserve our digital self with indistinguishable characteristics long after our physical presence ends. 

This work has also already begun. In China, services allow families to create AI avatars of deceased relatives using their photos, voice recordings, and social media history (The Guardian). Academic research such as Beyond Life: A Digital Will Solution for Posthumous Data Management outlines how digital assets could support controlled continuation of one’s digital presence after death (arXiv).

Happy eternity to you!