#openai发布gpt-6astra
When you look at the market and the direction it’s trading, every step AI takes forward raises the requirements for the underlying infrastructure by another level.
In the past, the market mainly focused on:
GPUs, computing power, and data centers.
But after the models continue to upgrade, the real bottleneck may not be only in chips.
Data throughput, memory capacity, and storage space will also become issues that AI expansion must solve.
So my view is:
In the second half of the year, it’s worth paying particular attention to the storage sector.
The AI main theme hasn’t ended; it may gradually shift from a purely “compute narrative” and spread to:
storage, high-bandwidth memory, and data center network infrastructure.
Especially if AI capital expenditures continue to grow, and if storage demand and industry profitability can keep up, the market may be able to re-price this direction.
AI won’t always just hype one kind of GPU.
The larger the model and the more data, the higher the inference demand—and the storage behind it becomes increasingly important.
Of course, topic hype doesn’t equal trend confirmation.
Going forward, we still need to see whether orders, supply and demand, performance, and trading volume can align.
My understanding is:
AI may need to return to the main track, while storage may be a branch worth observing early in the second half of the year.
In the second half of the year, I hope storage can return to the main line
When you look at the market and the direction it’s trading, every step AI takes forward raises the requirements for the underlying infrastructure by another level.
In the past, the market mainly focused on:
GPUs, computing power, and data centers.
But after the models continue to upgrade, the real bottleneck may not be only in chips.
Data throughput, memory capacity, and storage space will also become issues that AI expansion must solve.
So my view is:
In the second half of the year, it’s worth paying particular attention to the storage sector.
The AI main theme hasn’t ended; it may gradually shift from a purely “compute narrative” and spread to:
storage, high-bandwidth memory, and data center network infrastructure.
Especially if AI capital expenditures continue to grow, and if storage demand and industry profitability can keep up, the market may be able to re-price this direction.
AI won’t always just hype one kind of GPU.
The larger the model and the more data, the higher the inference demand—and the storage behind it becomes increasingly important.
Of course, topic hype doesn’t equal trend confirmation.
Going forward, we still need to see whether orders, supply and demand, performance, and trading volume can align.
My understanding is:
AI may need to return to the main track, while storage may be a branch worth observing early in the second half of the year.
In the second half of the year, I hope storage can return to the main line