美股盯盘日记
美股盯盘日记
美股盯盘，盯盘美股 持续更新，把握机会 新增关注：韩股、港股等
310Following
142followers
Feed
Feed
#SpaceX上市后首份完整季度财报将揭晓
A conspiracy theory: Is Musk playing a big game recently?
Looking at recent events together, it's quite interesting.
Two main threads:
· SpaceX and Tesla merger?
· Musk diving deeper into 🇺🇸 politics again?
① Technical collaboration → Merger expectations
October 15: Tesla reintroduced the second-generation Roadster, highlighting technical cooperation with SpaceX.
Mid-October: Barclays suggested that Tesla and SpaceX might merge by the end of the year, integrating AI, robotics, energy, and aerospace businesses.
② Capital layout → Political influence
October 19: Musk proposed a plan to gift SpaceX stock to children across the U.S., involving amounts possibly reaching tens of billions of dollars.
Coincidentally, this plan is linked to the U.S. government's Trump Accounts children's investment account policy.
More interestingly, according to media reports, Musk discussed related arrangements with the Trump administration as early as this summer.
③ October 20 to 21: Earnings report relay
SpaceX and Tesla successively released their Q3 earnings reports.
Connecting these events:
Technical collaboration → Merger expectations → Public shareholding → Earnings release
On one side, the business landscape may be heading toward integration; on the other, building interest ties with more American families through equity.
Is Musk hyping up the earnings report, or preparing to extend his business influence further into the political realm?
#特斯拉交付超预期后盈利修复成焦点
While domestic car companies are still debating whether a brake pedal should break or not, Tesla has already evolved into a "flying machine."
The moon is indeed rounder abroad.
Concepts can fly, but financial reports still need to land.
After the market closes on October 21, Tesla will release its Q3 financial report.
The previously disclosed delivery data was quite good.
Q3 deliveries reached 486,500 vehicles, production was 464,400 vehicles, and energy storage deployment hit 13.7 GWh.
The delivery volume has already given the market a good answer; now it depends on whether profits can keep up.
After all, selling more cars is one thing, but making more money is another.
If this time the automotive gross margin rebounds, and net profit and cash flow improve simultaneously, then Tesla's financial report will be even more valuable.
Deliveries have exceeded expectations; next, let's see if Tesla can deliver a financial report that also exceeds expectations.
#台积电Q3净利润超预期，盈利创历史新高
TSMC, the future is so bright it blinds me👀
It used to be said that selling shovels in the AI era was the most profitable, but now it seems the real money might be in making the shovels.
This earnings report is a bit exaggerated:
· Q3 revenue of 1.49 trillion New Taiwan Dollars, a year-on-year increase of 50.9%
· Net profit of 763.06 billion New Taiwan Dollars, a year-on-year increase of 68.7%, setting a record high
· Gross margin of 66.4%, and net margin reaching 51.1%
For every 100 units of revenue earned, more than 51 units become net profit
Major tech giants are competing upfront in AI and chip performance, while TSMC quietly makes money behind the scenes.
More importantly, TSMC not only benefits from the AI dividend but also controls the core capacity of advanced processes and advanced packaging.
· Q4 revenue guidance further raised to 52.3 billion–54.3 billion USD, indicating growth may not be over yet
However, the overall market fell today; earnings growth and stock price performance do not always move in sync.
Short-term market sentiment is hard to predict, but as long as AI giants continue to invest, TSMC, which controls core capacity, remains one of the most stable beneficiaries in this AI race
TSMC, the money printer of the AI era
#台积电Q3营收创新高，盈利表现待揭晓
It is estimated that this time it won't be bad either
Revenue data:
Q3 about 1.49 trillion New Taiwan Dollars, nearly 50% year-on-year growth, a historic high
Compared to chip competitors like Nvidia and AMD, TSMC as the manufacturer is the biggest beneficiary
Moreover, TSMC's own advantages are very obvious
Advanced process + advanced packaging
AI chip demand is also very strong, and capacity and yield directly affect how many chips customers can get
This time, focus on whether the gross margin can reach the previous guidance of 65% to 67%, and whether management will continue to increase investment in advanced processes and packaging
Companies that control core capacity are more worthy of long-term attention
ASML, making money quietly
The Q3 earnings report will be released on October 14
#阿斯麦Q3财报10月14日公布，EUV订单成焦点
Usually not as popular as Nvidia, but in the AI chip industry chain, its ability to make money is really strong
Currently, Q3 revenue is expected to be between 11 billion and 12 billion euros, with a gross margin of 55% to 57%
This profit margin is already quite competitive in manufacturing
This earnings result should still be very good
AI chip demand is very strong, and big manufacturers like TSMC and Samsung want to expand advanced process capacity, which cannot bypass ASML's EUV lithography machines
Another thing to pay attention to is how many new orders there are; buying equipment is one thing, but whether customers are willing to continue investing to expand production is also very important
As long as EUV orders remain strong, ASML still has great prospects in this AI capital expenditure cycle
#美联储官员密集发声，加息还要持续多久？
CME: The probability of a 25 basis point rate hike in October is about 54.2%, with a 45.8% chance of keeping rates unchanged.
It is estimated that a rate hike in October is unlikely.
The recent statements from Federal Reserve officials are quite interesting:
Barkin: Inflation risks still outweigh employment risks, but open to a few more hikes
Collins: Supports a September rate hike, inflation risk staying above 2% is rising
Musallam: Further tightening may still be needed
Kashkari: Inflationary pressures are no longer just from oil prices
Goolsbee: Relatively cautious about consecutive tightening
On one hand, inflation hasn't been subdued yet; on the other, there is policy pressure after consecutive tightening.
Raise or pause
Please choose your shadow hero
Oracle's "Three Highs," Do Not Disturb
The earnings report was just released last week, with revenue at 19.3 billion and cloud infrastructure up 121%, but the stock price has continued to decline over the past week.
Taking a closer look, wow: quarterly capital expenditure is 12 billion, free cash flow is negative 5 billion, debt-to-equity ratio is 500%, credit default swap spreads have hit the highest level since 2009, and the bond market has already priced it as junk.
A typical "Three Highs" target: high risk, high volatility, high potential return.
High risk: $50 billion annual capital expenditure, all-in on AI infrastructure, Oracle is a latecomer with a very short margin for error. More critically, about 50% of RPO is tied to OpenAI alone, while OpenAI's own annualized revenue is only 20 billion, but infrastructure commitments reach as high as 1.4 trillion—
High volatility: Q1 clearly exceeded expectations, yet the stock price still dropped 14% in a week because the market is not focused on revenue but on the financing narrative. Nearly 30,000 layoffs, with all the saved money poured into data centers, and traditional business still down 3% year-over-year.
High potential return: If AI infrastructure is truly a trillion-level demand as management claims, and if Oracle can hold on until data centers come online and RPO monetizes, TIKR's neutral model calculates a 2031 target price of $511. Looking back from today's $149, it might be the floor price.
#财报观察员：甲骨文AI云收入增121%
Storage Price Hike Shockwave: What Is the Mobile Phone Industry Going Through
After tonight, smartphone manufacturers may officially enter the era of 10,000 yuan. According to reports, Apple's foldable phones may start at $2,000, which is about 14,000 yuan.
The share of storage in mobile phone material costs has risen from 10%–15% to 30%–40%. Terminal brands can only raise prices or squeeze profits, and control costs by cutting procurement and specifications of non-core components, with upstream small and medium-sized manufacturers bearing the brunt.
Storage manufacturers have benefited significantly: Samsung Electronics' Q2 operating profit more than doubled for the whole of last year; SK Hynix's DRAM and NAND average prices rose about 30% and 55% quarter-on-quarter, respectively; The HBM market size is expected to grow 58% to $54.6 billion by 2026.
Prices are unlikely to fall in the short term. TrendForce expects Q3 DRAM and NAND contract prices to continue rising by 13%–18% and 10%–15%. Since most of the new capacity is directed to HBM, the incremental supply from consumer grade is limited, and real capacity release may not be released until the second half of 2027 or 2028.
This market is similar to the mining boom and global chip shortage, but AI demand is more sustainable, with new capacity prioritizing AI services. The essence is not ordinary cyclical fluctuations, but rather the structural squeeze of consumer electronics capacity by AI, and the era of "cheap and big-bodied phones" may be coming to an end.
#AI需求升温, Samsung SK Hynix's inventory is less than 10 days
Everyone, please stand up.
美股盯盘日记
Next, the tech industry will see three consecutive "Spring Festival Galas"
On September 7, Xiaomi held a launch event; on September 9, Apple held a launch event; on September 17, Huawei Connectivity Conference.
It seems like new terminal products are appearing one after another, but the real main theme is just one thing: chips.
Apple is betting on advanced processes. The iPhone 18 Pro is expected to debut the A20 Pro chip using TSMC's 2nm process, further enhancing performance and energy efficiency through GAA transistors.
Huawei chose to use architecture to supplement processes. According to the previously announced Ascend roadmap, the Ascend 950 will use self-developed HBM, with interconnect bandwidth 2.5 times higher than the 910C. With single-chip limitations, the company seeks incremental growth in storage, connectivity, and cluster architecture.
Xiaomi's focus is on the Xuanjie 03. The market is not only concerned about new products, but also whether self-developed chips can continue to iterate and enter more core products.
The three companies have different strategies, yet face the same problem: chips are becoming more expensive.
Manufacturers like China Resources Microelectronics and Infineon are raising prices intensively, with copper, wafer foundry, and packaging and testing costs rising simultaneously, and tight storage supply may persist until 2027.
Terminal price increases may seem to be happening at the counter, but the root cause is actually upstream.
New products will go viral, chips will set the upper limit. Process breakthroughs, architectural innovation, and supply chain restructuring will be the real main themes of the tech world in 2026
$xXIAOMI $xAAPL
GPT-6 Astra released: AGI is underway?
OpenAI released GPT-6 Astra last week, followed by Jensen Huang's high-profile announcement on social media, "AGI has arrived."
Discussions about Astra have surged like never before, and this time, the public statement is: it has achieved what previous models could not
From "answering questions" to "completing tasks"
*According to users who have already experienced it, it's really easy to use
But aside from the excellent experience, there's another internal debate at OpenAI: Does this count as AGI?
President Brockman hinted that "people looking back in the future might think AGI started from this point," but CEO Sam Altman had previously publicly stated that AGI was "vaguely defined" and "an irrelevant marketing term."
OpenAI defines it as "highly autonomous systems that surpass humans in the most economically valuable work."
This standard is neither high nor low, leaving plenty of room for ambiguity
Beyond this issue, what's even more troubling is that OpenAI itself admits that Astra "is more likely to intentionally hide or disguise its reasoning process," making it harder for humans to evaluate its decisions. There have also been incidents of models breaching test environments and infiltrating external systems.
Astra is truly a milestone advancement. When AI begins to act autonomously and learns hidden reasoning, will the world really be ready?
#日韩芯片股走强, can the AI storage cycle be extended?
