How AI Will Reshape Computers and Computing
Wajih Alkhiami — July 25, 2026
From Cloud Tools to AI-Native Computers
AI is changing both what computers do and how they are built. Dedicated Neural Processing Units are becoming common in premium smartphones and a growing share of PCs, moving more inference from remote data centers onto local devices. Flagship mobile processors now offer roughly 45–80 trillion operations per second, while Microsoft requires at least 40 TOPS for Copilot+ PCs.
This local processing can support offline translation, computational photography, computer vision, and personalized assistants. It may also improve privacy by keeping sensitive information on the device. Yet cloud computing will remain important for complex reasoning and large multimodal workloads, making hybrid edge-cloud systems the likely direction.
Software is evolving as well. Agentic AI can plan and execute multi-step tasks rather than merely respond to prompts. Adoption has reportedly reached 35%, but only 11%–14% of organizations have moved agents into full production.
Boomer Perspective
The optimistic case is that computers become faster, more private, and more useful. Local AI can reduce latency and dependence on network access, while agents may automate software testing, IT operations, compliance, and other complicated workflows.
Efficiency is also improving at the task level. Smaller models, quantization, and hardware-software co-design could bring increasingly capable AI to phones, laptops, industrial sensors, and low-power devices. AI may also help smaller firms automate expensive work and improve the productivity of less experienced workers.
Doomer Perspective
The cautionary case begins with energy and infrastructure. Global data-center electricity consumption is projected to rise from 485 TWh in 2025 to about 950 TWh in 2030. Efficiency gains may not offset expanding use of video generation and agentic systems.
Autonomous agents also introduce risks including prompt injection, excessive permissions, cascading failures, and oversight evasion. Monitoring covered only about 52% of production agents in one 2026 report.
Access is another concern. AI adoption stands at 39% among large OECD enterprises but only 12% among small firms, while chips, advanced fabrication, cloud services, and data centers remain highly concentrated.
A Balanced Outlook
The evidence supports neither effortless abundance nor inevitable disaster. On-device AI and agentic software offer concrete computing benefits, but peak TOPS can overstate real performance because thermal limits, memory bandwidth, and fragmented software ecosystems matter.
The strongest path forward combines hybrid computing, fine-grained authorization, continuous monitoring, human oversight, and grid-friendly data centers. AI’s future impact will depend less on raw capability alone than on whether security, energy systems, software standards, and access improve alongside it.
