What the future looks like
In five to ten years, much of AI runs on the device in front of you or close to it. Cars, robots, glasses and PCs make split-second decisions locally, because they can't wait for a round trip to a distant data center, and private data can stay on the device.
The cloud still trains the biggest models, but inference spreads to billions of endpoints. Each one needs its own AI chip and plenty of fast memory.
How big could it get?
- Global spending on edge computing is about $261 billion in 2025, rising to about $380 billion by 2028, a 13.8% annual growth rate (IDC, 2025).
- IDC notes that AI-accelerated processors are driving hardware spending early in the forecast, with demand from agentic AI at the edge.
Figures are third-party estimates, quoted as published. Different studies use different scopes and years, so they aren't always comparable.
What we're watching
- AI PCs and desktop AI supercomputers reaching mainstream buyers.
- Sales of AI glasses and other always-on assistants.
- More capable in-car and in-robot computers.
- Memory content per device (DRAM and storage), a key driver for memory makers.
What could go wrong
- Devices are cyclical and price-sensitive; consumer upgrade cycles can stall.
- Smaller on-device models may stay much less capable than cloud models.
- Fragmented hardware makes it hard for developers to build once and run everywhere.
- Memory pricing is volatile through the cycle.
What the future makers say
“Every software engineer, every engineer, every creative artist — everybody who uses computers today as a tool — will need an AI supercomputer.”
“Personal devices like glasses … will become our primary computing devices.”
Quotes are reproduced from the linked public sources; paraphrases are labeled. Inclusion doesn't imply any affiliation with or endorsement by these people or companies.
Companies building it
Only companies we own. Tickers only, never position sizes.
In-car DRIVE computers, robot computers, RTX AI PCs and the Project DIGITS desktop AI supercomputer (NVIDIA, 2025).
DRAM and storage for phones, PCs, cars and edge servers; more AI per device means more memory per device.
AI glasses that see and hear what you do, which it expects to become a primary computing device.
Runs its driving AI on its own in-vehicle computer, and the same approach powers Optimus.
Sources
Last reviewed: 22 Oct 2026.
Not investment advice. This page is opinion and education, researched and presented with the help of AI analyst personas, and may contain errors. The EarlyConviction team holds positions in NVDA, MU, META, TSLA and may buy or sell at any time. Read the disclaimer.