yoyo777's Profile | Binance Square

·

--

$SKHYNIX did some homework 📄 First, Meta has not cut back on AI investment—if anything, it’s continuing to ramp up. According to the latest disclosed information, Meta’s 2026 AI infrastructure (CapEx) budget is about $145 billion, one of the highest levels in history. The company hasn’t said it plans to reduce GPU purchases or scale back data center construction. But what the market is worried about is whether the investment can start making money quickly enough. First, what did Zuckerberg say? At a recent internal employee meeting, Mark Zuckerberg admitted: The R&D progress on AI agents is slower than originally expected; For AI, the company has already made large-scale organizational changes, moving thousands of employees to AI teams; He expects it will take another 3 to 6 months before the company starts seeing more noticeable results. The market’s interpretation is: After spending so much money, there hasn’t been corresponding commercial return in the short term. Second, why can one sentence make AI chip stocks drop? Let’s assume the market’s original expectation was: Meta spends $145 billion → buys more GPUs → GPUs get more HBM → TSMC continues to reap huge profits. Now it becomes: Meta still spends $145 billion, but the deployment speed of AI applications is slower than expected, so will procurement pace over the next two or three years be not as fast as before? Note: This isn’t “demand disappears,” but “growth rate may slow down.” For AI stocks whose valuations are already very high, if the growth rate drops from 100% to 50%, the stock price could see a major correction. Third, why is there news again about “selling AI compute”? Recently, there were also reports that Meta is studying ways to commercialize some of its AI compute power and provide AI compute services to external customers. The market has produced two completely different interpretations: The optimistic camp believes: Meta has spent so much money, and now it’s finally starting to take these GPUs and monetize them. This suggests the investment is entering the return stage. The pessimistic camp believes: If its own GPUs aren’t even enough, why would it still sell to others? Does that mean: Internal demand isn’t as high as expected? Some GPU capacity is idle? AI data center construction might be moving too fast? Right now, the market seems to lean toward the second explanation, so the AI hardware sector is under pressure.

$SKHYNIX did some homework 📄
First, Meta has not cut back on AI investment—if anything, it’s continuing to ramp up.
According to the latest disclosed information, Meta’s 2026 AI infrastructure (CapEx) budget is about $145 billion, one of the highest levels in history. The company hasn’t said it plans to reduce GPU purchases or scale back data center construction.
But what the market is worried about is whether the investment can start making money quickly enough.
First, what did Zuckerberg say?
At a recent internal employee meeting, Mark Zuckerberg admitted:
The R&D progress on AI agents is slower than originally expected;
For AI, the company has already made large-scale organizational changes, moving thousands of employees to AI teams;
He expects it will take another 3 to 6 months before the company starts seeing more noticeable results.
The market’s interpretation is:
After spending so much money, there hasn’t been corresponding commercial return in the short term.

Second, why can one sentence make AI chip stocks drop?
Let’s assume the market’s original expectation was:
Meta spends $145 billion → buys more GPUs → GPUs get more HBM → TSMC continues to reap huge profits.
Now it becomes:
Meta still spends $145 billion,
but the deployment speed of AI applications is slower than expected,
so will procurement pace over the next two or three years be not as fast as before?
Note:
This isn’t “demand disappears,” but “growth rate may slow down.”
For AI stocks whose valuations are already very high,
if the growth rate drops from 100% to 50%,
the stock price could see a major correction.

Third, why is there news again about “selling AI compute”?
Recently, there were also reports that Meta is studying ways to commercialize some of its AI compute power and provide AI compute services to external customers.
The market has produced two completely different interpretations:
The optimistic camp believes:
Meta has spent so much money, and now it’s finally starting to take these GPUs and monetize them.
This suggests the investment is entering the return stage.
The pessimistic camp believes:
If its own GPUs aren’t even enough, why would it still sell to others?
Does that mean:
Internal demand isn’t as high as expected?
Some GPU capacity is idle?
AI data center construction might be moving too fast?
Right now, the market seems to lean toward the second explanation, so the AI hardware sector is under pressure.

META+0.15%

SKHYNIX+0.65%

METAUS-4.86%

yoyo777

Trending Topics