#AI股持续上涨还有哪些投资机会
Don’t just stare at compute when doing AI. These few less-discussed directions in AI are nobody talks about—but there’s real money to be made.
Go to the public square now and it’s all GPUs, optical modules, and liquid cooling. It’s so competitive it makes your scalp tingle. That’s not to say these are bad. It’s just that too many people are there, so the excess returns have long been squeezed dry.
Let me share a few directions I find interesting, but almost nobody discusses 👇
1. Telecom networks
AI is moving from the “brain” to the “body.” Robots and autonomous driving need edge computing, and the network has to keep up. An Ericsson executive said the next winners might be telecom operators that lay fiber cables. BofA predicts that the AI network market will reach $316 billion by 2030. Domestic carriers have already started selling token packages—basically like selling data plans. This direction is practically zero-discussed in the square.
2. Multispectral AI
In simple terms, it means giving machines a wider range of vision than human eyes—covering infrared and ultraviolet as well—so temperature anomalies and arc hazards can be detected early. Policy is pushing it forward. By 2030, the market is expected to hit $79.4 billion, growing at 31.8% year over year. The top five players’ market share is only 10.9%—so there’s basically no “crowding.” Power grid inspections and IDC security are already using it.
3. Emotion AI
Not competing on doing work—compete on companionship. Pet emotion collars and smart plant sensors have already proven willingness to pay. The parameter race for big models is a completely different dimension from this, but those businesses really do get paid.
CZ once said that AI × Crypto is the fastest path to real-world deployment. Between AI agents, small, frequent payments will likely use cryptocurrencies. All of the scenarios above are exactly what need micro-payments and pay-per-effect pricing. The more AI becomes practical, the more real the demand for crypto payments.
Risk warning: This is only my personal opinion and does not constitute investment advice.
Which one do you like more? Comment below
Don’t just stare at compute when doing AI. These few less-discussed directions in AI are nobody talks about—but there’s real money to be made.
Go to the public square now and it’s all GPUs, optical modules, and liquid cooling. It’s so competitive it makes your scalp tingle. That’s not to say these are bad. It’s just that too many people are there, so the excess returns have long been squeezed dry.
Let me share a few directions I find interesting, but almost nobody discusses 👇
1. Telecom networks
AI is moving from the “brain” to the “body.” Robots and autonomous driving need edge computing, and the network has to keep up. An Ericsson executive said the next winners might be telecom operators that lay fiber cables. BofA predicts that the AI network market will reach $316 billion by 2030. Domestic carriers have already started selling token packages—basically like selling data plans. This direction is practically zero-discussed in the square.
2. Multispectral AI
In simple terms, it means giving machines a wider range of vision than human eyes—covering infrared and ultraviolet as well—so temperature anomalies and arc hazards can be detected early. Policy is pushing it forward. By 2030, the market is expected to hit $79.4 billion, growing at 31.8% year over year. The top five players’ market share is only 10.9%—so there’s basically no “crowding.” Power grid inspections and IDC security are already using it.
3. Emotion AI
Not competing on doing work—compete on companionship. Pet emotion collars and smart plant sensors have already proven willingness to pay. The parameter race for big models is a completely different dimension from this, but those businesses really do get paid.
CZ once said that AI × Crypto is the fastest path to real-world deployment. Between AI agents, small, frequent payments will likely use cryptocurrencies. All of the scenarios above are exactly what need micro-payments and pay-per-effect pricing. The more AI becomes practical, the more real the demand for crypto payments.
Risk warning: This is only my personal opinion and does not constitute investment advice.
Which one do you like more? Comment below