叮当 Doraemon
173 Posts
叮当 Doraemon
Square VerifiedReport
Block User
5年炒币亏成狗，2026埋头苦干web3
514 Following
32.1K+ Followers
34.4K+ Liked
Posts
All
Quotes
Videos
PINNED
·
--
Article
Musk admits: Terafab is in talks with TSMC 🏭Previously, when everyone looked at Terafab, there was at least some sense of “is Musk trying to take over everything himself again?” After all, advanced process nodes aren’t like building rockets, and spending money doesn’t automatically mean you can get it done.
The wind direction has changed a bit now: even TSMC is researching how to get involved.
Culpan’s take is that a more realistic option could be: TSMC builds the factories itself and operates them itself in Texas. SpaceX / Terafab would provide the funding, lock in long-term orders, and even hold shares together.
That is to say, Terafab may not ultimately really be “Musk himself making chips.” More likely, it becomes “Musk puts the demand and the money on the table and invites the best chip makers to do it.”
Musk admits: Terafab is in talks with TSMC 🏭
Previously, when everyone looked at Terafab, there was at least some sense of “is Musk trying to take over everything himself again?” After all, advanced process nodes aren’t like building rockets, and spending money doesn’t automatically mean you can get it done.
The wind direction has changed a bit now: even TSMC is researching how to get involved.
Culpan’s take is that a more realistic option could be: TSMC builds the factories itself and operates them itself in Texas. SpaceX / Terafab would provide the funding, lock in long-term orders, and even hold shares together.
That is to say, Terafab may not ultimately really be “Musk himself making chips.” More likely, it becomes “Musk puts the demand and the money on the table and invites the best chip makers to do it.”
The wind direction has changed a bit now: even TSMC is researching how to get involved.
Culpan’s take is that a more realistic option could be: TSMC builds the factories itself and operates them itself in Texas. SpaceX / Terafab would provide the funding, lock in long-term orders, and even hold shares together.
That is to say, Terafab may not ultimately really be “Musk himself making chips.” More likely, it becomes “Musk puts the demand and the money on the table and invites the best chip makers to do it.”
2.4k
14
123
2.4k
PINNED
·
--
Verified
Article
The semiconductor ETF portfolio of $AMZN is very likely to end up being more attractive than what $NVDA holds……Three days ago, Amazon signed a $49.8 million private placement investment, injecting funds into PCB manufacturer Gold Circuit Electronics (2368). So now Amazon’s stake/ warrants include:
- $AAOI
- Alchip
- $MRVL (Celestial)
- $ALAB
- $CRDO
- $JBL
- $FLEX
- $STM
- $CBRS
- $GNRC
- $FN
- TD SYNNEX
- $QCOM
- Gold Circuit Electronics
and others……
If you look specifically at GCE’s private investment amount in the context of the deal, it is equivalent to 0.36% of the company…… financially it’s a bit small, but strategically it’s meaningful,
In particular, there are claims that GCE provides ASIC server PCBs to the four largest hyperscale CSPs……
The semiconductor ETF portfolio of $AMZN is very likely to end up being more attractive than what $NVDA holds……
Three days ago, Amazon signed a $49.8 million private placement investment, injecting funds into PCB manufacturer Gold Circuit Electronics (2368). So now Amazon’s stake/ warrants include:
- $AAOI
- Alchip
- $MRVL (Celestial)
- $ALAB
- $CRDO
- $JBL
- $FLEX
- $STM
- $CBRS
- $GNRC
- $FN
- TD SYNNEX
- $QCOM
- Gold Circuit Electronics
and others……
If you look specifically at GCE’s private investment amount in the context of the deal, it is equivalent to 0.36% of the company…… financially it’s a bit small, but strategically it’s meaningful,
In particular, there are claims that GCE provides ASIC server PCBs to the four largest hyperscale CSPs……
- $AAOI
- Alchip
- $MRVL (Celestial)
- $ALAB
- $CRDO
- $JBL
- $FLEX
- $STM
- $CBRS
- $GNRC
- $FN
- TD SYNNEX
- $QCOM
- Gold Circuit Electronics
and others……
If you look specifically at GCE’s private investment amount in the context of the deal, it is equivalent to 0.36% of the company…… financially it’s a bit small, but strategically it’s meaningful,
In particular, there are claims that GCE provides ASIC server PCBs to the four largest hyperscale CSPs……
2.5k
11
141
2.5k
·
--
Verified
Article
Morgan Stanley Reinstates NVIDIA as Its Top Semiconductor Pick: Industry Bottlenecks Shift to Data Center Infrastructure, with Growth Opportunities from Collaborations with SpaceX and AmazonMorgan Stanley puts NVIDIA back at the top of its semiconductor picks: industry bottlenecks are shifting to data center infrastructure, with partnerships involving SpaceX and Amazon opening up growth opportunities
Morgan Stanley’s latest research report has once again named chip giant $NVIDIA (NVDA.US)$ as its top pick in the semiconductor sector, reiterating a “Buy” rating and setting a target price of $300. The firm believes industry bottlenecks are rapidly shifting from semiconductor production capacity constraints to the pace of new data center construction and financing models. NVIDIA, it says, is well positioned to capture the industry tailwinds thanks to its product architecture, global customer ecosystem, and financing support capabilities. The report also points to potential incremental upside from Muse intelligent agents and new customer collaborations, arguing that the company’s valuation has room to move higher. Morgan Stanley further emphasized that NVIDIA’s valuation, based on its FY2028 earnings forecast, implies a price-to-earnings (P/E) ratio of just 15x, making it highly attractive. Even if the P/E multiple does not expand, solid investment returns can still be achieved through earnings growth.
Morgan Stanley Reinstates NVIDIA as Its Top Semiconductor Pick: Industry Bottlenecks Shift to Data Center Infrastructure, with Growth Opportunities from Collaborations with SpaceX and Amazon
Morgan Stanley puts NVIDIA back at the top of its semiconductor picks: industry bottlenecks are shifting to data center infrastructure, with partnerships involving SpaceX and Amazon opening up growth opportunities
Morgan Stanley’s latest research report has once again named chip giant $NVIDIA (NVDA.US)$ as its top pick in the semiconductor sector, reiterating a “Buy” rating and setting a target price of $300. The firm believes industry bottlenecks are rapidly shifting from semiconductor production capacity constraints to the pace of new data center construction and financing models. NVIDIA, it says, is well positioned to capture the industry tailwinds thanks to its product architecture, global customer ecosystem, and financing support capabilities. The report also points to potential incremental upside from Muse intelligent agents and new customer collaborations, arguing that the company’s valuation has room to move higher. Morgan Stanley further emphasized that NVIDIA’s valuation, based on its FY2028 earnings forecast, implies a price-to-earnings (P/E) ratio of just 15x, making it highly attractive. Even if the P/E multiple does not expand, solid investment returns can still be achieved through earnings growth.
Morgan Stanley’s latest research report has once again named chip giant $NVIDIA (NVDA.US)$ as its top pick in the semiconductor sector, reiterating a “Buy” rating and setting a target price of $300. The firm believes industry bottlenecks are rapidly shifting from semiconductor production capacity constraints to the pace of new data center construction and financing models. NVIDIA, it says, is well positioned to capture the industry tailwinds thanks to its product architecture, global customer ecosystem, and financing support capabilities. The report also points to potential incremental upside from Muse intelligent agents and new customer collaborations, arguing that the company’s valuation has room to move higher. Morgan Stanley further emphasized that NVIDIA’s valuation, based on its FY2028 earnings forecast, implies a price-to-earnings (P/E) ratio of just 15x, making it highly attractive. Even if the P/E multiple does not expand, solid investment returns can still be achieved through earnings growth.
AMZNUS-0.08%
NVDAUS+0.68%
SPCXUS+0.81%
1
413
·
--
Oh wow! SpaceX’s vertical integration is moving so fast—it’s basically doing “everything it makes itself.”
Latest channel reports say SpaceX is already considering building its own fiber-optic production capacity. In fact, fiber cabling has been produced internally for some time. The fiber they’re using today mainly comes from Corning, Hengtong, and FiberHome. But as Starlink expands together with AI data centers, the demand to replace the supply chain will noticeably rise by 2027.
This makes perfect sense in the context of the 10GW compute target: millions of GPUs ultimately need high-speed optical interconnects. If any link—fibers or optical modules—gets bottlenecked, the entire scale-up has to wait.
Musk used to use this playbook when building rockets: if a component might bottleneck the supply chain, he brings it in-house. Now this approach is being replicated from rockets and satellites all the way to AI data centers.
#spcx
Oh wow! SpaceX’s vertical integration is moving so fast—it’s basically doing “everything it makes itself.”
Latest channel reports say SpaceX is already considering building its own fiber-optic production capacity. In fact, fiber cabling has been produced internally for some time. The fiber they’re using today mainly comes from Corning, Hengtong, and FiberHome. But as Starlink expands together with AI data centers, the demand to replace the supply chain will noticeably rise by 2027.
This makes perfect sense in the context of the 10GW compute target: millions of GPUs ultimately need high-speed optical interconnects. If any link—fibers or optical modules—gets bottlenecked, the entire scale-up has to wait.
Musk used to use this playbook when building rockets: if a component might bottleneck the supply chain, he brings it in-house. Now this approach is being replicated from rockets and satellites all the way to AI data centers.
#spcx
Latest channel reports say SpaceX is already considering building its own fiber-optic production capacity. In fact, fiber cabling has been produced internally for some time. The fiber they’re using today mainly comes from Corning, Hengtong, and FiberHome. But as Starlink expands together with AI data centers, the demand to replace the supply chain will noticeably rise by 2027.
This makes perfect sense in the context of the 10GW compute target: millions of GPUs ultimately need high-speed optical interconnects. If any link—fibers or optical modules—gets bottlenecked, the entire scale-up has to wait.
Musk used to use this playbook when building rockets: if a component might bottleneck the supply chain, he brings it in-house. Now this approach is being replicated from rockets and satellites all the way to AI data centers.
#spcx
1
446
·
--
Bullish
He who understands the times is a true hero!
In the face of power, even Old Ma has to bow his head.
Just a few days ago, Trump issued an order: from now on, the U.S. federal government should not call it AI anymore—use SI instead (Super Intelligence).
Old Ma immediately jumped in to praise it: “No more AI. SI. It’s better.” Then today someone asked: “Should SpaceXAI just rename itself to SpaceXSI?” Old Ma replied right away: “Yes, we will make that change.” And they actually changed it.
You should know that xAI only just merged into SpaceX, and it hasn’t been long since they changed it to SpaceXAI—now they’re making a second renaming.
Trump comes up with the new name, and Old Ma demonstrates his support with the first real action. This round of flattery even puts the company’s signboard in the same frame.
Although I have reservations about this kind of approach, not going against the president probably won’t put them at a disadvantage, right?
$SPCX
He who understands the times is a true hero!
In the face of power, even Old Ma has to bow his head.
Just a few days ago, Trump issued an order: from now on, the U.S. federal government should not call it AI anymore—use SI instead (Super Intelligence).
Old Ma immediately jumped in to praise it: “No more AI. SI. It’s better.” Then today someone asked: “Should SpaceXAI just rename itself to SpaceXSI?” Old Ma replied right away: “Yes, we will make that change.” And they actually changed it.
You should know that xAI only just merged into SpaceX, and it hasn’t been long since they changed it to SpaceXAI—now they’re making a second renaming.
Trump comes up with the new name, and Old Ma demonstrates his support with the first real action. This round of flattery even puts the company’s signboard in the same frame.
Although I have reservations about this kind of approach, not going against the president probably won’t put them at a disadvantage, right?
$SPCX
In the face of power, even Old Ma has to bow his head.
Just a few days ago, Trump issued an order: from now on, the U.S. federal government should not call it AI anymore—use SI instead (Super Intelligence).
Old Ma immediately jumped in to praise it: “No more AI. SI. It’s better.” Then today someone asked: “Should SpaceXAI just rename itself to SpaceXSI?” Old Ma replied right away: “Yes, we will make that change.” And they actually changed it.
You should know that xAI only just merged into SpaceX, and it hasn’t been long since they changed it to SpaceXAI—now they’re making a second renaming.
Trump comes up with the new name, and Old Ma demonstrates his support with the first real action. This round of flattery even puts the company’s signboard in the same frame.
Although I have reservations about this kind of approach, not going against the president probably won’t put them at a disadvantage, right?
$SPCX
1
1.6k
·
--
Verified
Rather than focusing on the ups and downs of AI tech leaders, I personally prefer to dig into companies’ potential in the practical applications of fiber-optic templates.
Potential FCC rules on optical transceivers are more likely to target 3.2T — $LITE, $AAOI, $COHR, $SIVE.
Morgan Stanley meets with Washington officials:
- Restrictions are likely to start with 3.2T modules manufactured in China (800G/1.6T will not be affected)
- Potential restrictions are likely to keep the laser market tight (in combination with the U.S. DSP + laser possibly reaching MS’s 65% U.S. value threshold per item)
- If the 65% bill of materials is U.S. content, China-made modules may qualify. China modules using U.S. DSP and lasers are already close to that level
- Analysts expect the positive impact from laser pricing… to be more important than any broad reallocation of market share among optical transceivers.
- Analysts do not expect this to become a bottleneck for hyperscale AI buildouts; they point out that the key variable to watch is the supply of InP substrates (hello $AXTI).
TLDR:
Component suppliers overall are very optimistic. Names like $SMTC could be winners for TIA/drivers (I don’t think it’s listed), and maybe $MXL for DSP as well.
$AAOI is another major beneficiary.
$SIVE is high-potential alpha for laser beneficiaries because they have “huge capacity” coming online, pointing to U.S. contract manufacturers (they have not officially disclosed the locations yet).
Because previous slides showed 2 U.S. contract manufacturers and 1 Taiwan contract manufacturer (Win Semi) as capacity partners. And this is based on elimination.
From checking the CIOE channel, they have indeed started working with China plug-and-play players (so as beneficiaries of components in the U.S. supply chain, they have a path forward).
Of course $LITE / $COHR will be happy.
TLDR: Laser / DSP / TIA / other West-side alternative chains are already bottlenecked, but qualified capacity is becoming even more important.
Rather than focusing on the ups and downs of AI tech leaders, I personally prefer to dig into companies’ potential in the practical applications of fiber-optic templates.
Potential FCC rules on optical transceivers are more likely to target 3.2T — $LITE, $AAOI, $COHR, $SIVE.
Morgan Stanley meets with Washington officials:
- Restrictions are likely to start with 3.2T modules manufactured in China (800G/1.6T will not be affected)
- Potential restrictions are likely to keep the laser market tight (in combination with the U.S. DSP + laser possibly reaching MS’s 65% U.S. value threshold per item)
- If the 65% bill of materials is U.S. content, China-made modules may qualify. China modules using U.S. DSP and lasers are already close to that level
- Analysts expect the positive impact from laser pricing… to be more important than any broad reallocation of market share among optical transceivers.
- Analysts do not expect this to become a bottleneck for hyperscale AI buildouts; they point out that the key variable to watch is the supply of InP substrates (hello $AXTI).
TLDR:
Component suppliers overall are very optimistic. Names like $SMTC could be winners for TIA/drivers (I don’t think it’s listed), and maybe $MXL for DSP as well.
$AAOI is another major beneficiary.
$SIVE is high-potential alpha for laser beneficiaries because they have “huge capacity” coming online, pointing to U.S. contract manufacturers (they have not officially disclosed the locations yet).
Because previous slides showed 2 U.S. contract manufacturers and 1 Taiwan contract manufacturer (Win Semi) as capacity partners. And this is based on elimination.
From checking the CIOE channel, they have indeed started working with China plug-and-play players (so as beneficiaries of components in the U.S. supply chain, they have a path forward).
Of course $LITE / $COHR will be happy.
TLDR: Laser / DSP / TIA / other West-side alternative chains are already bottlenecked, but qualified capacity is becoming even more important.
Potential FCC rules on optical transceivers are more likely to target 3.2T — $LITE, $AAOI, $COHR, $SIVE.
Morgan Stanley meets with Washington officials:
- Restrictions are likely to start with 3.2T modules manufactured in China (800G/1.6T will not be affected)
- Potential restrictions are likely to keep the laser market tight (in combination with the U.S. DSP + laser possibly reaching MS’s 65% U.S. value threshold per item)
- If the 65% bill of materials is U.S. content, China-made modules may qualify. China modules using U.S. DSP and lasers are already close to that level
- Analysts expect the positive impact from laser pricing… to be more important than any broad reallocation of market share among optical transceivers.
- Analysts do not expect this to become a bottleneck for hyperscale AI buildouts; they point out that the key variable to watch is the supply of InP substrates (hello $AXTI).
TLDR:
Component suppliers overall are very optimistic. Names like $SMTC could be winners for TIA/drivers (I don’t think it’s listed), and maybe $MXL for DSP as well.
$AAOI is another major beneficiary.
$SIVE is high-potential alpha for laser beneficiaries because they have “huge capacity” coming online, pointing to U.S. contract manufacturers (they have not officially disclosed the locations yet).
Because previous slides showed 2 U.S. contract manufacturers and 1 Taiwan contract manufacturer (Win Semi) as capacity partners. And this is based on elimination.
From checking the CIOE channel, they have indeed started working with China plug-and-play players (so as beneficiaries of components in the U.S. supply chain, they have a path forward).
Of course $LITE / $COHR will be happy.
TLDR: Laser / DSP / TIA / other West-side alternative chains are already bottlenecked, but qualified capacity is becoming even more important.
1
1.6k
·
--
Bullish
Latest Anthropic IPO filing disclosures: signed a maximum $84.5 billion NVIDIA compute power agreement with SpaceX, covering up to 2029. $SPCX
The agreement can be canceled with a 90-day notice; it’s part of its decade-long $51.8 billion infrastructure plan. Previously, it had already been connected to Colossus with more than 220,000 GPUs and over 300 MW of production capacity.
This reflects that AI compute capacity has become the core bottleneck, and SpaceX data centers are shifting from internal use for xAI to large-scale external leasing.
#spacex
Latest Anthropic IPO filing disclosures: signed a maximum $84.5 billion NVIDIA compute power agreement with SpaceX, covering up to 2029. $SPCX
The agreement can be canceled with a 90-day notice; it’s part of its decade-long $51.8 billion infrastructure plan. Previously, it had already been connected to Colossus with more than 220,000 GPUs and over 300 MW of production capacity.
This reflects that AI compute capacity has become the core bottleneck, and SpaceX data centers are shifting from internal use for xAI to large-scale external leasing.
#spacex
The agreement can be canceled with a 90-day notice; it’s part of its decade-long $51.8 billion infrastructure plan. Previously, it had already been connected to Colossus with more than 220,000 GPUs and over 300 MW of production capacity.
This reflects that AI compute capacity has become the core bottleneck, and SpaceX data centers are shifting from internal use for xAI to large-scale external leasing.
#spacex
4
964
·
--
Space X is the most valuable company of the next 50 years
Worth building a position in batches and holding
SpaceX Starlink delivers stable income.
Plus selling computing power to it and Google—now selling two to three hundred billion in computing power per year, and the payback period has been compressed to within a year. So Claude is worth two trillion. Based on model inference, SpaceX is at least 5 trillion, even 10 trillion.
In the future, Starship transportation will be fully operational.
Just think about the resources on the Moon and Mars
Think about space travel
Think about satellite communications
Mm~ damn 🔥
Purely my personal opinion and does not constitute investment advice
Space X is the most valuable company of the next 50 years
Worth building a position in batches and holding
SpaceX Starlink delivers stable income.
Plus selling computing power to it and Google—now selling two to three hundred billion in computing power per year, and the payback period has been compressed to within a year. So Claude is worth two trillion. Based on model inference, SpaceX is at least 5 trillion, even 10 trillion.
In the future, Starship transportation will be fully operational.
Just think about the resources on the Moon and Mars
Think about space travel
Think about satellite communications
Mm~ damn 🔥
Purely my personal opinion and does not constitute investment advice
Worth building a position in batches and holding
SpaceX Starlink delivers stable income.
Plus selling computing power to it and Google—now selling two to three hundred billion in computing power per year, and the payback period has been compressed to within a year. So Claude is worth two trillion. Based on model inference, SpaceX is at least 5 trillion, even 10 trillion.
In the future, Starship transportation will be fully operational.
Just think about the resources on the Moon and Mars
Think about space travel
Think about satellite communications
Mm~ damn 🔥
Purely my personal opinion and does not constitute investment advice
1.4k
12
94
1.4k
·
--
Article
When AGI completes everything for humans, what can we still do? Philosophy provides the answer, and trading will become the practice of testing that answer.Many people worry about AI because they fear losing their jobs.
Programmers worry that AI will write code, designers worry that AI will make drawings, doctors worry that AI will diagnose diseases, analysts worry that AI will read financial reports, and fund managers worry that AI will directly complete stock selection and trading.
But this isn’t the most frightening part yet.
The real question is:
Suppose one day, AGI will not only replace our physical labor, but also replace most knowledge, skills, judgment, and creativity—what will humans have left?
When AI can read more than you, calculate faster, remember more firmly, and even understand your habits better than you do, the abilities you once took pride in may no longer be scarce.
When AGI completes everything for humans, what can we still do? Philosophy provides the answer, and trading will become the practice of testing that answer.
Many people worry about AI because they fear losing their jobs.
Programmers worry that AI will write code, designers worry that AI will make drawings, doctors worry that AI will diagnose diseases, analysts worry that AI will read financial reports, and fund managers worry that AI will directly complete stock selection and trading.
But this isn’t the most frightening part yet.
The real question is:
Suppose one day, AGI will not only replace our physical labor, but also replace most knowledge, skills, judgment, and creativity—what will humans have left?
When AI can read more than you, calculate faster, remember more firmly, and even understand your habits better than you do, the abilities you once took pride in may no longer be scarce.
Programmers worry that AI will write code, designers worry that AI will make drawings, doctors worry that AI will diagnose diseases, analysts worry that AI will read financial reports, and fund managers worry that AI will directly complete stock selection and trading.
But this isn’t the most frightening part yet.
The real question is:
Suppose one day, AGI will not only replace our physical labor, but also replace most knowledge, skills, judgment, and creativity—what will humans have left?
When AI can read more than you, calculate faster, remember more firmly, and even understand your habits better than you do, the abilities you once took pride in may no longer be scarce.
2
4
14
626
·
--
Verified
Article
Backed by solid evidence of confidence! Nvidia adds $150 billion to its buyback, setting a record for total authorized repurchases by a U.S. companyAs it continues to reap massive profits from its top position in the AI supply chain, Nvidia announced an additional $150 billion stock buyback program, setting a record for the largest buyback in U.S. corporate history.
The company, which has the highest market value globally, said on Monday that its board of directors has approved an increase of $150 billion to its share repurchase program. Together with the existing plan, Nvidia currently has authorization to deploy up to $235 billion for stock buybacks by January 2028. Buoyed by the news, Nvidia’s share price rose more than 3% in early trading.
NVDA Nvidia
228.820
Backed by solid evidence of confidence! Nvidia adds $150 billion to its buyback, setting a record for total authorized repurchases by a U.S. company
As it continues to reap massive profits from its top position in the AI supply chain, Nvidia announced an additional $150 billion stock buyback program, setting a record for the largest buyback in U.S. corporate history.
The company, which has the highest market value globally, said on Monday that its board of directors has approved an increase of $150 billion to its share repurchase program. Together with the existing plan, Nvidia currently has authorization to deploy up to $235 billion for stock buybacks by January 2028. Buoyed by the news, Nvidia’s share price rose more than 3% in early trading.
NVDA Nvidia
228.820
The company, which has the highest market value globally, said on Monday that its board of directors has approved an increase of $150 billion to its share repurchase program. Together with the existing plan, Nvidia currently has authorization to deploy up to $235 billion for stock buybacks by January 2028. Buoyed by the news, Nvidia’s share price rose more than 3% in early trading.
NVDA Nvidia
228.820
2.1k
15
120
2.1k
·
--
Article
The second half of AI profit distribution: who’s winning, who’s still undecided, and who’s about to be disruptedNow’s AI and tomorrow’s AI
If we divide the AI industry chain into upstream equipment, midstream cloud providers, and large model companies, then the downstream is application companies. Currently, almost all the profits in the industry are concentrated in the upstream. For example, just Nvidia (NVDA.US) plus the four major storage manufacturers take nearly 70% of the industry’s profits.
This kind of profit distribution shows the typical characteristics of the early stage of a new technology revolution. But let’s think: in the future, the charging entry points for the AI industry won’t be cloud providers or downstream application companies—so will profit distribution still look like this?
Taking a mature internet industry as an example, the approximate proportions of economic profit distribution are:
The second half of AI profit distribution: who’s winning, who’s still undecided, and who’s about to be disrupted
Now’s AI and tomorrow’s AI
If we divide the AI industry chain into upstream equipment, midstream cloud providers, and large model companies, then the downstream is application companies. Currently, almost all the profits in the industry are concentrated in the upstream. For example, just Nvidia (NVDA.US) plus the four major storage manufacturers take nearly 70% of the industry’s profits.
This kind of profit distribution shows the typical characteristics of the early stage of a new technology revolution. But let’s think: in the future, the charging entry points for the AI industry won’t be cloud providers or downstream application companies—so will profit distribution still look like this?
Taking a mature internet industry as an example, the approximate proportions of economic profit distribution are:
If we divide the AI industry chain into upstream equipment, midstream cloud providers, and large model companies, then the downstream is application companies. Currently, almost all the profits in the industry are concentrated in the upstream. For example, just Nvidia (NVDA.US) plus the four major storage manufacturers take nearly 70% of the industry’s profits.
This kind of profit distribution shows the typical characteristics of the early stage of a new technology revolution. But let’s think: in the future, the charging entry points for the AI industry won’t be cloud providers or downstream application companies—so will profit distribution still look like this?
Taking a mature internet industry as an example, the approximate proportions of economic profit distribution are:
2.5k
17
163
2.5k
·
--
Lao Huang truly lives up to the title of the universe’s No. 1 mass-marketing pitchman, putting on a big show together with the universe’s No. 1 president.
What did Huang Renxun say are the advantages of AI in the United States? It’s Donald Trump!
To be honest, many things Trump does are to pander to populism, but when it comes to grand strategy, he’s absolutely not sloppy. The only way for the United States of America to maintain its competitive edge is to concentrate resources to take on big tasks—policies and resources (including capital) must escort the building of AI infrastructure.
The “exception” story about the beautiful country is never outdated!
Last night, when Huang Renxun was going All-in and delivering a speech on stage to thousands, Trump called in. Huang put the call on speaker so the whole room could hear the conversation. Trump said: The AI doomsday risk is a hoax, and the argument against building data centers is also a hoax. He insists that they cannot let opposition in some places prevent the expansion of America’s computing-power infrastructure. A really badass live counterattack performance art—during last night’s intraday trading, the U.S. stock market recovered a lot of losses. This phone call certainly made no small contribution.
Lao Huang truly lives up to the title of the universe’s No. 1 mass-marketing pitchman, putting on a big show together with the universe’s No. 1 president.
What did Huang Renxun say are the advantages of AI in the United States? It’s Donald Trump!
To be honest, many things Trump does are to pander to populism, but when it comes to grand strategy, he’s absolutely not sloppy. The only way for the United States of America to maintain its competitive edge is to concentrate resources to take on big tasks—policies and resources (including capital) must escort the building of AI infrastructure.
The “exception” story about the beautiful country is never outdated!
Last night, when Huang Renxun was going All-in and delivering a speech on stage to thousands, Trump called in. Huang put the call on speaker so the whole room could hear the conversation. Trump said: The AI doomsday risk is a hoax, and the argument against building data centers is also a hoax. He insists that they cannot let opposition in some places prevent the expansion of America’s computing-power infrastructure. A really badass live counterattack performance art—during last night’s intraday trading, the U.S. stock market recovered a lot of losses. This phone call certainly made no small contribution.
What did Huang Renxun say are the advantages of AI in the United States? It’s Donald Trump!
To be honest, many things Trump does are to pander to populism, but when it comes to grand strategy, he’s absolutely not sloppy. The only way for the United States of America to maintain its competitive edge is to concentrate resources to take on big tasks—policies and resources (including capital) must escort the building of AI infrastructure.
The “exception” story about the beautiful country is never outdated!
Last night, when Huang Renxun was going All-in and delivering a speech on stage to thousands, Trump called in. Huang put the call on speaker so the whole room could hear the conversation. Trump said: The AI doomsday risk is a hoax, and the argument against building data centers is also a hoax. He insists that they cannot let opposition in some places prevent the expansion of America’s computing-power infrastructure. A really badass live counterattack performance art—during last night’s intraday trading, the U.S. stock market recovered a lot of losses. This phone call certainly made no small contribution.
1
475
·
--
Elon Musk on how to make AI safer and truly for people:
“AI will be smarter than the smartest human. At that point, any invention will become possible.”
But AI also has a small chance of killing all of us.
So the most important thing is how we train it.
• Make AI as realistic as possible
• Make it maximally curious
• Train it to stay honest even if the truth isn’t popular
Because if AI truly pursues the truth and is full of curiosity, he believes it will naturally want to promote human development, rather than fight against it.
Elon Musk on how to make AI safer and truly for people:
“AI will be smarter than the smartest human. At that point, any invention will become possible.”
But AI also has a small chance of killing all of us.
So the most important thing is how we train it.
• Make AI as realistic as possible
• Make it maximally curious
• Train it to stay honest even if the truth isn’t popular
Because if AI truly pursues the truth and is full of curiosity, he believes it will naturally want to promote human development, rather than fight against it.
“AI will be smarter than the smartest human. At that point, any invention will become possible.”
But AI also has a small chance of killing all of us.
So the most important thing is how we train it.
• Make AI as realistic as possible
• Make it maximally curious
• Train it to stay honest even if the truth isn’t popular
Because if AI truly pursues the truth and is full of curiosity, he believes it will naturally want to promote human development, rather than fight against it.
630
·
--
Article
Oh no, it’s over—everyone goes their own way. A trader lost more than $190,000 from chasing LAPTOPSeptember 9, according to Lookonchain monitoring, a trader chased higher prices for a token related to Hunter Biden, LAPTOP, and lost $200,000, leaving only about $3,000. The trader withdrew $250,000 from Binance in advance, planning to buy immediately after LAPTOP goes live. In the end, he spent $200,000 to buy 919 LAPTOP at a high price of $218 per token; currently, this position is worth only about $3,000.
Oh no, it’s over—everyone goes their own way. A trader lost more than $190,000 from chasing LAPTOP
September 9, according to Lookonchain monitoring, a trader chased higher prices for a token related to Hunter Biden, LAPTOP, and lost $200,000, leaving only about $3,000. The trader withdrew $250,000 from Binance in advance, planning to buy immediately after LAPTOP goes live. In the end, he spent $200,000 to buy 919 LAPTOP at a high price of $218 per token; currently, this position is worth only about $3,000.
6
2
10
2.1k
·
--
Article
Bitcoin May See a “Golden Cross” as a Decline in the USDT Share Releases a Risk-On SignalOn September 3, Bitcoin is currently nearing the formation of a “golden cross” technical pattern in which the 50-day moving average crosses above the 200-day moving average. This indicator is typically viewed by the market as a signal of a long-term uptrend.
Although historical performance doesn’t always accurately predict market movements, this golden cross may also receive further support from a decline in the USDT share, which could signal a shift toward higher risk appetite.
In Bitcoin’s history, there have been 12 golden crosses in total, and some stages have brought significant upward market rallies.
Statistics show that among 9 measurable golden crosses, the 3-month average gain is about 24.9%. While there are fewer cases where the trend was sustained for a full year without being broken by a death cross, the average gain over a one-year period for the previous 3 cases reached 250%.
Bitcoin May See a “Golden Cross” as a Decline in the USDT Share Releases a Risk-On Signal
On September 3, Bitcoin is currently nearing the formation of a “golden cross” technical pattern in which the 50-day moving average crosses above the 200-day moving average. This indicator is typically viewed by the market as a signal of a long-term uptrend.
Although historical performance doesn’t always accurately predict market movements, this golden cross may also receive further support from a decline in the USDT share, which could signal a shift toward higher risk appetite.
In Bitcoin’s history, there have been 12 golden crosses in total, and some stages have brought significant upward market rallies.
Statistics show that among 9 measurable golden crosses, the 3-month average gain is about 24.9%. While there are fewer cases where the trend was sustained for a full year without being broken by a death cross, the average gain over a one-year period for the previous 3 cases reached 250%.
Although historical performance doesn’t always accurately predict market movements, this golden cross may also receive further support from a decline in the USDT share, which could signal a shift toward higher risk appetite.
In Bitcoin’s history, there have been 12 golden crosses in total, and some stages have brought significant upward market rallies.
Statistics show that among 9 measurable golden crosses, the 3-month average gain is about 24.9%. While there are fewer cases where the trend was sustained for a full year without being broken by a death cross, the average gain over a one-year period for the previous 3 cases reached 250%.
1
380
·
--
Real business success isn’t about how much money you make, but how much warmth you bring to others🩷
Real business success isn’t about how much money you make, but how much warmth you bring to others🩷
3.8k
17
255
3.8k
·
--
Article
The $200 Billion Financing Monster Behind Anthropic: Wall Street Buys the Chips, With Google and Broadcom Providing BackstopsAI Flash News: Behind Anthropic’s Google computing power order, a financing and contracting system worth about $200 billion has already been put in place. (The Financial Times) says that roughly 80% of it is related to TPU chips. Anthropic does not need to fork out all the money upfront; the chip costs are mainly funded by Wall Street, while Google and Broadcom respectively back the data centers and the chips.
The first batch—about $35 billion, 1GW of TPU—was initially sold by Google to Broadcom, then transferred to an SPV (a financing company set up specifically for the deal). Institutions such as Apollo and Blackstone provide debt financing to the SPV, which buys about 1 million TPU chips, and then leases them to Anthropic.
The $200 Billion Financing Monster Behind Anthropic: Wall Street Buys the Chips, With Google and Broadcom Providing Backstops
AI Flash News: Behind Anthropic’s Google computing power order, a financing and contracting system worth about $200 billion has already been put in place. (The Financial Times) says that roughly 80% of it is related to TPU chips. Anthropic does not need to fork out all the money upfront; the chip costs are mainly funded by Wall Street, while Google and Broadcom respectively back the data centers and the chips.
The first batch—about $35 billion, 1GW of TPU—was initially sold by Google to Broadcom, then transferred to an SPV (a financing company set up specifically for the deal). Institutions such as Apollo and Blackstone provide debt financing to the SPV, which buys about 1 million TPU chips, and then leases them to Anthropic.
The first batch—about $35 billion, 1GW of TPU—was initially sold by Google to Broadcom, then transferred to an SPV (a financing company set up specifically for the deal). Institutions such as Apollo and Blackstone provide debt financing to the SPV, which buys about 1 million TPU chips, and then leases them to Anthropic.
323
·
--
Article
The world’s third-largest Bitcoin company megawhale
Asia's "MicroStrategy" Metaplanet transferred another 1,350 BTC to Coinbase Prime one hour ago, about $108 millionOn August 28, according to Lookonchain monitoring, the Bitcoin treasury company Metaplanet transferred another 1,350 BTC to Coinbase Prime about one hour ago, worth about $108 million.
Previously, the company bought 43,000 BTC at an average price of $96,191, worth approximately $3.48 billion.
Metaplanet announced on August 18 that it would inject 2,100 BTC into the Nasdaq-listed company Super League to build a U.S. stock treasury platform. Just ten days later, it again transferred 1,350 BTC to Coinbase Prime. This was the second large transfer within the past week (it transferred 1,000 BTC on August 25), and the pace is clearly accelerating in crypto.
The world’s third-largest Bitcoin company megawhale Asia's "MicroStrategy" Metaplanet transferred another 1,350 BTC to Coinbase Prime one hour ago, about $108 million
On August 28, according to Lookonchain monitoring, the Bitcoin treasury company Metaplanet transferred another 1,350 BTC to Coinbase Prime about one hour ago, worth about $108 million.
Previously, the company bought 43,000 BTC at an average price of $96,191, worth approximately $3.48 billion.
Metaplanet announced on August 18 that it would inject 2,100 BTC into the Nasdaq-listed company Super League to build a U.S. stock treasury platform. Just ten days later, it again transferred 1,350 BTC to Coinbase Prime. This was the second large transfer within the past week (it transferred 1,000 BTC on August 25), and the pace is clearly accelerating in crypto.
Previously, the company bought 43,000 BTC at an average price of $96,191, worth approximately $3.48 billion.
Metaplanet announced on August 18 that it would inject 2,100 BTC into the Nasdaq-listed company Super League to build a U.S. stock treasury platform. Just ten days later, it again transferred 1,350 BTC to Coinbase Prime. This was the second large transfer within the past week (it transferred 1,000 BTC on August 25), and the pace is clearly accelerating in crypto.
1
684
·
--
Better Strategy Than Skill? Let Sincerity Prevail
If a person is smooth in every direction, eloquent, and especially good at “playing the game,” it doesn’t necessarily mean they have high emotional intelligence.
Real high emotional intelligence is: once this person steps forward in front of others, they are trusted endlessly. They are also good at sharing good things with others. When others are in trouble, they are willing to help. With this kind of emotional intelligence, among the crowd they will never be rejected.
Being worldly, good at “being in the right place,” and mastering various methods and techniques of social interaction are only at the level of “technique” (shu). The level of “the Way” (dao) is character—how one treats others by putting oneself in their shoes, guided by virtue.
Face all complexity with pure intention, and use your truest heart and actions to weed out those who can’t resonate with you.
Better Strategy Than Skill? Let Sincerity Prevail
If a person is smooth in every direction, eloquent, and especially good at “playing the game,” it doesn’t necessarily mean they have high emotional intelligence.
Real high emotional intelligence is: once this person steps forward in front of others, they are trusted endlessly. They are also good at sharing good things with others. When others are in trouble, they are willing to help. With this kind of emotional intelligence, among the crowd they will never be rejected.
Being worldly, good at “being in the right place,” and mastering various methods and techniques of social interaction are only at the level of “technique” (shu). The level of “the Way” (dao) is character—how one treats others by putting oneself in their shoes, guided by virtue.
Face all complexity with pure intention, and use your truest heart and actions to weed out those who can’t resonate with you.
If a person is smooth in every direction, eloquent, and especially good at “playing the game,” it doesn’t necessarily mean they have high emotional intelligence.
Real high emotional intelligence is: once this person steps forward in front of others, they are trusted endlessly. They are also good at sharing good things with others. When others are in trouble, they are willing to help. With this kind of emotional intelligence, among the crowd they will never be rejected.
Being worldly, good at “being in the right place,” and mastering various methods and techniques of social interaction are only at the level of “technique” (shu). The level of “the Way” (dao) is character—how one treats others by putting oneself in their shoes, guided by virtue.
Face all complexity with pure intention, and use your truest heart and actions to weed out those who can’t resonate with you.
11
1.6k
·
--
Niúlái and Pénglái fight, a bit looking forward to it
Niúlái and Pénglái fight, a bit looking forward to it
4
4
12
3.9k
Log in to explore more content
Join global crypto users on Binance Square
⚡️ Get latest and useful information about crypto.
💬 Trusted by the world’s largest crypto exchange.
👍 Discover real insights from verified creators.
Email / Phone number
Trending Topics
DriftHackVictimsBeginClaims
24,976 views
423 Discussing
🚨 $DRIFT HACK VICTIMS CAN NOW START CLAIMING
Victims of the Drift hack are beginning to receive access to the claims process.
After months of uncertainty, affected users finally have a path toward recovering their losses. 💰
If you were impacted, check the official claim details carefully and avoid fake links or scammers. ⚠️
$SOL
#DRIFT #solana #DriftHackVictimsBeginClaims $1000000BOB
Adeem Jutt
·1 Likes
·99 views
SolanaTokenizedStockVolumeTops$4.4BInSeptember
4,093 views
50 Discussing
BitcoinSpotETFsDraw$6.34BInflowsInQ3
14,763 views
162 Discussing