纳指100是信仰

$SPCX First technical adjustment, time to stack up!

Why am I still bullish on $SPCX ?

Actually, SPCX's three major business lines create a flywheel effect.
SpaceX isn't just a single rocket company; it's three interlocking sectors:

1. Launch Services: Completing 165 orbital launches by 2025, with 500 milestones for Falcon 9. The cost per launch has dropped from the industry standard of $400 million to $62 million. The Falcon 9 booster has achieved 29 successful reuses with a success rate over 99%. ​​

2. Starlink Satellite Internet: Low latency (25-60 ms), high speed (up to 215 Mbps), far surpassing traditional geostationary satellites at about 20 Mbps. Laser links create a mesh network, with each satellite serving up to 4,000 users. Terminal costs have fallen from $3,000 at launch to $600. ​​

3. Starship Next-Gen Rocket: The FAA has authorized increasing Starbase's launch frequency from 5 to 25 times a year. If Starship achieves scalable operations, launch costs will drop another order of magnitude, unlocking orbital freight, space tourism, and deep space missions. ​This trio of businesses forms a self-reinforcing flywheel: launch revenues fund R&D, cheaper rockets deploy more satellites, more satellites bring in more Starlink users and revenue, and more revenue is reinvested into next-gen technologies.

In the short term, passive buying from index inclusion and underwriter reports may drive the stock price higher. In the medium term, September earnings and Cursor integration progress are critical validation points. Long-term, if any of Starship's commercialization, AI satellite constellations, or Terafab chip factories succeed, the current valuation will be redefined.

Short term target of $220-250 (corresponding to index inclusion and report catalysts), medium term target of $300-350 (corresponding to AI business ramp-up and profitability inflection), and a long-term target of over $500 (corresponding to the realization of the platform vision).

$SPCX picking up passengers 👇

Wall Street Analysts Turn Bullish on CRDO $CRDO

Recently, CRDO has seen a surge in analyst coverage and upward ratings:

Evercore ISI: Initiated coverage with an outperform rating and a target price of $325. Analyst Mark Lipacis believes the market is underestimating Credo's optical potential, forecasting an EPS exceeding $13 by 2028, 40% higher than Wall Street's consensus.
​​
Stifel: Maintained a buy rating, raising the target price from $250 to $350. The management's two-day roadshow conveyed the core strategy of "vertical integration and system-level solutions."
​​
Out of 18 analysts, 16 are rating it a buy, 1 a strong buy, only 2 hold, and zero sell. The consensus rating is a "Moderate Buy," with a short-term target of $350.

$WDC is picking someone up

1. Technical Barriers
• Hard drive manufacturing is an extremely capital-intensive industry, with only 3 major players left globally (Western Digital, Seagate, Samsung), making it nearly impossible for new entrants to break in
• Western Digital has a deep accumulation of storage density technologies like HAMR (Heat-Assisted Magnetic Recording) and SMR (Shingled Magnetic Recording)

2. Economies of Scale
• Building a factory costs tens of billions of dollars, and Western Digital dilutes costs through large-scale production, making it tough for smaller players to compete
• After merging with SanDisk, it covers both Hard Disk Drives (HDD) and Solid State Drives (SSD), providing a complete product line

3. Customer Stickiness
• Once data centers and cloud computing companies (AWS, Azure, Google) select a supplier, the switching costs are extremely high—requiring re-validation of compatibility and stability, which takes a long time
• Enterprise storage demands high reliability, and customers tend to favor long-term partnerships

4. Supply Chain Control
• The upstream supply chain for core components like read/write heads and platters is highly concentrated, and Western Digital locks in key resources through vertical integration and long-term agreements

In summary: This isn’t a moat built on some "black tech," but rather a combination of "high capital barriers + technological accumulation + customer lock-in" that forms an oligopoly; it’s tough for outsiders to replicate no matter how much money they throw at it.

Welcome to join the long position army 👇

$NBIS is about to break above 300

$SKHYNIX has AI core HBM: a high-demand, long-term play, with a market projection of $45 billion by 2026 and a compound annual growth rate of 33% from 2025 to 2030. The supply-demand gap is expected to last at least until the end of 2027; plus, SK Hynix holds a 58% market share in HBM, ranking first globally, and is a key supplier to Nvidia, maximizing the sector's profit potential.

Short-term (2026-2028) growth is assured:

1. HBM4E samples have been sent to Nvidia for the next-gen GPUs, with single-card capacity continually on the rise;

2. HBM capacity as a percentage of DRAM wafer production has increased from 18% to 30%, with high-margin products making up a growing share;

Welcome to the long side 👇

$MRVL Market close dumped seven points

Reverse trade incoming

Opportunity knocks, don't miss out

Join the bulls👇

$MRVL short term target 400, long term target 1000

Welcome to the bulls

$CRDO 's moat

In a nutshell: CRDO's moat isn't about tech monopoly; it's a trifecta of "first-mover advantage + customer lock-in + patent walls" that’s only getting wider.


What's it all about?
In AI data centers, thousands of GPUs need to chat at high speed. CRDO’s AEC (Active Electrical Cable) serves as the lifeline between them—copper cables packed with chips to ensure signals travel further, more stably, and with less power consumption.

The core selling point is singular: ZeroFlap. Any drop in connection during AI training could waste hours of computing power and electricity. This cable is pricey, but worth it.


Where's the moat?
First, 15 years of honing SerDes craftsmanship. High-speed analog circuit design can't just be bought; it demands deep knowledge of silicon processes and a mountain of real-world data. CRDO has been grinding on this since 2008; while $MRVL has the cash, in this niche, $CRDO is the one that got there first.

Second, customer certification lock-in. Once CRDO's AEC is "certified" into a data center's rack configuration, switching suppliers means a complete system re-certification—firmware, management software, rack layout, all starting over. What's even tougher is that its PILOT platform is already embedded in the customer's switch management system; it’s not just a cable, it’s a "self-diagnosing nervous system".

Third, the most elegant move: turning competitors into "rent payers". Giants like InnoLight and Luxshare tried to make their own AECs, but CRDO went straight to ITC for a patent battle, resulting in those three rivals becoming its licensed partners—they can produce but must pay CRDO patent fees. Turning foes into tenants is the highest form of moat.

Fourth, it invented the AEC category and currently holds about 88% market share. In AI cluster connectivity solutions, regular copper cables are cheap but signal-poor, while optical modules are pricey and prone to dropouts. AEC hits the sweet spot of "moderate distance, affordable, stable, low power". CRDO’s not just selling products; it's defining the standards for that sweet spot.

There won't be another 300 under $MRVL anymore.

Marvell (美满电子) is a US fabless chip designer, focusing on AI computational interconnect chips: their main offerings include high-end optical DSPs, high-speed switching chips, and custom AI chips for cloud providers, alongside telecom and automotive networking chips; the optical communication high-end chips are dominated by a duopoly with Broadcom, deeply tied to Nvidia and top global cloud providers.

There's ample room for growth in the mid to long term, positioning itself in the high-demand "high-speed connectivity" lane, and optical chips have nearly monopolistic barriers, making them a core asset for AI computational infrastructure.

$QQQ , time to swap out five companies; the era of wired tech is over, now it's all about AI and rocket science.

This contract is a bit sketchy for going long, but if I can snag some gains from the US stock contracts, I'll grab some spot QQQ and just HODL.

$ANTHROPIC

Just how strong does something have to be to get banned from trading?

I believe the opportunity is right here.

$ANTHROPIC This company, at its core, is all about one word: stability.

1. The founders come from OpenAI, but their philosophy is even stronger.

Dario and Daniela Amodei, siblings and former key executives at OpenAI, left to go solo not for the cash, but because they felt OpenAI's commercialization was too crazy and safety research wasn't keeping up. So from day one, Anthropic set a hard rule: safety first, profits later. This kind of "twist" has actually attracted top-tier research talent.

2. Money without sacrificing safety; investors must sign a "safety agreement."

Google and Amazon are eager to invest, but Anthropic has a strict condition: investors can't interfere with the tech roadmap and must support their safety research. While other startups would practically kneel before their investors, Anthropic stands tall and earns money—while making sure the investors follow their pace.

3. The team density is absurdly high.

Because of their strong safety philosophy, Anthropic has become the "holy land" for AI safety researchers. Talent isn't coming for the high salaries; they come because "we're really doing the right thing here." This kind of values-driven cohesion is extremely rare in Silicon Valley.

4. Not chasing trends, they have their own rhythm.

While others are fixated on model parameters and user growth, Anthropic is quietly focusing on alignment research, long-form text, and constitutional AI. It may seem slow in the short term, but every step is solid. The capital markets actually trust this kind of long-term certainty more.

Anthropic may not be the "biggest" AI company, but it's definitely the "toughest"—making money while choosing its funding, and uniting top talent through values. In Silicon Valley, having this kind of confidence is worth more than the technology itself.

This aligns perfectly with the goals of value investing 👇

One of the big three super companies' stock surged to over $200 after its IPO, $SPCX .

So, will the next opportunity be with the remaining two?

$OPENAI $ANTHROPIC

Time to stack a small position early and hold tight after the listing👇

I recently revisited Mavenir $MRVL , and the more I look into it, the more I feel like this company is undervalued by many.

In AI servers, GPUs handle the computation, but how is data transmitted? How do servers connect? How do thousands of GPUs work in sync? If these issues aren't solved properly, no matter how strong the computational power is, it won't be utilized effectively.

What Mavenir is doing is precisely these foundational infrastructures.

To put it simply, in the AI era, it's not just about computational power; transmission capability is key too. As GPUs get more powerful, data traffic increases, and the demand for high-speed networks and optical connections will only grow.

That's why I think the logic behind Mavenir is quite sound.

As AI develops faster, the network requirements increase; the higher the network demand, the more apparent Mavenir's value becomes.

It's not always easy to predict who will win the big model competition, but as long as people keep building data centers, buying servers, and expanding computational power, investment in foundational infrastructure is a must.

Often, the companies that really make big bucks aren't necessarily the ones that shine the brightest in the spotlight, but rather the ones that everyone can't avoid.

This is exactly why Huang points out that he could very well be the next trillion-dollar company.

Your uncle in the Nasdaq has seen clearly a lot of people this week.

We're almost back to the all-time highs now😍

No wonder it's a hotspot for pension funds from Japan, Korea, Canada, Australia, the US, the UK, China, and other countries🥵

The rewards center lets you farm interest for 500 on the $SXT for seven days, even if it's just pocket change.

A lot of folks think TSMC is already overpriced, but the real question is:

Who else in the world can replace it?

After the AI boom, the rarest commodity isn’t the models, but the advanced fabrication processes.

No matter how strong NVIDIA is, they still need TSMC for their foundry services;
Apple, no matter how big, can't escape TSMC's grasp.

The scariest part of the chip industry isn’t just being ahead; it’s being so far ahead that there are no competitors.

The market always worries about cycles, but TSMC isn’t just selling chips anymore; they’re selling the "shovels" of the global tech industry.

As long as AI keeps evolving, the demand for computational power won’t stop, and TSMC is the core toll gate of the computational era.

My take is pretty simple:

Buying TSMC is essentially a bet on the tech growth of the next decade.

$TSM

I recently took another look at $INTC, and I feel like the market is still pricing it a bit pessimistically, but its future potential is huge.

First off, the most important point is that Intel has been transforming from a traditional CPU company to one that does both design and foundry work. This is what they call IDM 2.0. Simply put, it’s not just about making chips for themselves anymore; they want to help others with chip production as well. If they nail this direction, they won't just be a 'follower' but could evolve into something like a fab platform company.

Secondly, the whole AI narrative has been a bit one-dimensional in the market. Everyone's focused on GPUs right now, but once AI truly rolls out, a lot of the work is actually done on the inference and edge side, where CPUs are still super important. Intel's stronghold in server CPUs and the PC market is still very much intact and massive; this part won't just be replaced overnight.

Thirdly, we've got the manufacturing process issue. In the past, the biggest fear was that Intel had fallen too far behind, but the reality now is that while the gap still exists, it’s not a complete chasm anymore. If their 18A nodes can progress as planned, the market’s confidence in their tech could very well bounce back.

Fourth, let's talk about the external environment. The semiconductor landscape is now more than just commercial competition; it’s leaning towards national strategy. The U.S. is pushing for domestic manufacturing, and clients don’t want to have their supply chains all in one basket. This is a big plus for Intel since it’s based in the U.S. and has full manufacturing capabilities.

Lastly, there’s the expectation issue. Intel’s biggest problem right now isn’t a weak foundation; it’s that the market has pre-emptively priced it as a loser. Once they achieve any milestone in foundry or process, there’s actually a lot of room for stock price and valuation growth.

Overall, it's not one of those high-certainty leaders; it’s more like a company at a critical transformation phase, so now is a prime opportunity to scoop up some cheap chips 👇

The market is completely blind to the value of $TSM . Who needs Nvidia or AMD? Samsung? Intel? Just give them three more years to catch up.

Still bullish on $TSM

Big players like Apple and NVIDIA aren't just buying its services, but rather:
They can only stabilize production using it.

Because the risk of switching suppliers is too high:
If the yield drops even a bit, the product crashes.
Time delays can make you miss the market window.
The cost of rebuilding the ecosystem is extremely high.
This creates a term: lock-in effect.

TSMC isn't just a cyclical stock or regular manufacturing; it’s the only stable exporter of high-end chips globally.
As AI develops, it feels more like "invisible infrastructure" rather than just a single company.

If you look at it from a trading perspective, its biggest advantage isn’t how fast it grows, but rather —
It's tough for anyone to knock it off the table.

The best time to plant a tree was ten years ago, the second best time is now👇