AI stocks by segment
Building AI infrastructure takes six different kinds of company, and they do not move together. Each group below is scored on its own, weighted by market capitalisation.
Prices as of 2026-10-03 00:39 UTC. Updates live in your browser.
Segment sentiment, last 16 sessions
Each line is one segment's cap-weighted market sentiment. Where the lines diverge, one part of the supply chain is repricing on its own; where they move together, it is a broad move in AI infrastructure and the individual scores tell you less.
3 Sept to 2 Oct. Each panel shares one 0-100 scale, with the other five segments behind it in grey.
Optical & networking has moved furthest over the window at +37, Memory & storage least at -13. Trading sessions only, so each step is one market day against the one before it.
AI chips
Companies designing the processors that train and run AI models - GPUs, custom accelerators and the CPUs alongside them.
Optical & networking
The interconnect layer. Once a model spans thousands of accelerators, the switches and optics between them become the bottleneck.
All 6 optical & networking stocks →
Memory & storage
High-bandwidth memory and the storage beneath it. Memory bandwidth, not compute, is what most often limits AI training throughput.
All 4 memory & storage stocks →
Neoclouds
Specialist GPU cloud operators renting accelerated compute, typically on multi-year contracts and heavily debt-funded.
Equipment & EDA
The tools and software that make chips possible - lithography, deposition, etch, inspection and the design software behind every layout.
All 7 equipment & eda stocks →
Power & systems
Racks, cooling and electricity. A datacentre full of accelerators is ultimately a power and thermal problem.
All 4 power & systems stocks →
Common questions
What are the segments of the AI chip supply chain?
AI chips, Optical & networking, Memory & storage, Neoclouds, Equipment & EDA, Power & systems. Together they cover 34 US-listed companies, from the processors themselves through to the electricity that runs them.
Which part of the AI complex has the strongest sentiment right now?
Equipment & EDA scores highest at 79 out of 100 (hot), and Memory & storage lowest at 42 (neutral). Segment scores are weighted by market capitalisation, so the largest names dominate the reading.
Why split AI stocks into segments at all?
Because they do not move together. Memory and equipment respond to capital-expenditure cycles that run years long; neoclouds respond to financing conditions and contract news; optics tracks the interconnect bottleneck. Grouping by what a company actually sells separates a broad move in AI infrastructure from one part of the chain repricing on its own.
Go deeper
Market data and information only, not investment advice.