What should AI be investing in now? These big industry sectors have great opportunities
Everyone knows about GPUs, and in this cycle the prices have already fully reflected it.
So what other industries have big opportunities in the next phase of AI?
I think mainly these:
1️⃣ Electricity
The more AI compute power there is, the more electricity it consumes—so power infrastructure becomes a hard necessity.
For example: nuclear power, gas-fired power generation, and grid equipment.
2️⃣ Networking and optical modules
The more GPUs there are, the greater the demand for data transmission within data centers. Networking and optical communications are a necessity.
For example: optical modules, switches.
3️⃣ Liquid cooling and data center heat dissipation
AI chips are consuming more and more power, and traditional air cooling is becoming increasingly difficult.
For example: liquid cooling systems, CDUs, cold plates.
4️⃣ AI servers
GPUs are just chips. To actually deploy compute power, you need the entire server infrastructure.
For example: AI servers, server ODMs.
5️⃣ AI software and services that truly help enterprises make money or save money
Hardware sells compute power, but software ultimately needs to prove ROI. Only AI applications that directly improve efficiency and reduce costs have the potential to produce the next batch of major companies.
For example: AI that truly replaces large amounts of repetitive work—AI customer service, AI programming, AI office work, and AI enterprise automation—starting to scale up commercial adoption.
Everyone knows about GPUs, and in this cycle the prices have already fully reflected it.
So what other industries have big opportunities in the next phase of AI?
I think mainly these:
1️⃣ Electricity
The more AI compute power there is, the more electricity it consumes—so power infrastructure becomes a hard necessity.
For example: nuclear power, gas-fired power generation, and grid equipment.
2️⃣ Networking and optical modules
The more GPUs there are, the greater the demand for data transmission within data centers. Networking and optical communications are a necessity.
For example: optical modules, switches.
3️⃣ Liquid cooling and data center heat dissipation
AI chips are consuming more and more power, and traditional air cooling is becoming increasingly difficult.
For example: liquid cooling systems, CDUs, cold plates.
4️⃣ AI servers
GPUs are just chips. To actually deploy compute power, you need the entire server infrastructure.
For example: AI servers, server ODMs.
5️⃣ AI software and services that truly help enterprises make money or save money
Hardware sells compute power, but software ultimately needs to prove ROI. Only AI applications that directly improve efficiency and reduce costs have the potential to produce the next batch of major companies.
For example: AI that truly replaces large amounts of repetitive work—AI customer service, AI programming, AI office work, and AI enterprise automation—starting to scale up commercial adoption.