#微软新笔电搭载英伟达ai芯片
👉 端侧AI怎么看，币安聊天室跟进
Microsoft has released a new Surface laptop.
Priced at $2,599.
Inside is an NVIDIA AI chip.
Shipping starts on October 16.
The focus isn’t really on the laptop itself.
It’s NVIDIA putting the chip into personal devices.
In the past, AI ran in the cloud.
Now compute power is being moved to the edge/terminal.
This step will change the boundaries for latency, cost, and privacy.
If a model can run locally, there’s no need to send it to the cloud.
And the data doesn’t have to leave home.
For chip makers, this is a new channel for shipping.
The PC upgrade cycle was already coming to an end.
AI has become a reason to restart the demand.
Vendors bundle the chips, the system, and the applications together and sell them as one.
The closer the entry point is to the front, the more stable the position.
The challenges are also very clear.
The price isn’t cheap; ordinary users may not buy into it.
What local models can do is still quite limited.
On-device/edge is only just getting started.
For encryption, terminal-side compute power is getting stronger.
The forms of wallets and proxies will also change.
Where do you think on-device AI will become mainstream first? Let’s discuss in the comments.
👉 端侧AI怎么看，币安聊天室跟进
Microsoft has released a new Surface laptop.
Priced at $2,599.
Inside is an NVIDIA AI chip.
Shipping starts on October 16.
The focus isn’t really on the laptop itself.
It’s NVIDIA putting the chip into personal devices.
In the past, AI ran in the cloud.
Now compute power is being moved to the edge/terminal.
This step will change the boundaries for latency, cost, and privacy.
If a model can run locally, there’s no need to send it to the cloud.
And the data doesn’t have to leave home.
For chip makers, this is a new channel for shipping.
The PC upgrade cycle was already coming to an end.
AI has become a reason to restart the demand.
Vendors bundle the chips, the system, and the applications together and sell them as one.
The closer the entry point is to the front, the more stable the position.
The challenges are also very clear.
The price isn’t cheap; ordinary users may not buy into it.
What local models can do is still quite limited.
On-device/edge is only just getting started.
For encryption, terminal-side compute power is getting stronger.
The forms of wallets and proxies will also change.
Where do you think on-device AI will become mainstream first? Let’s discuss in the comments.