美股投资young

美股科技投资者 ai狂热者分享每日快讯. all in ai

303Following

768followers

Feed

Livestream

Feed

Without Jensen Huang, faces great uncertainty in its prospects, and I think the negative impact is quite significant. But at the 160 price level, based on orders and profits, it is still seriously undervalued, far from my target price. I have shared the exact figure in the members' group. Undervaluation is one thing; discounting the outlook is another. Orders won't disappear just because the CEO changes. Capital expenditures from , , and are still increasing, and production capacity is booked through next year and the year after. Next week's earnings report will likely still look very good. But the greatness of has never been just about how many cards it sells. It doesn't make its own chips or server racks. Chips rely on , HBM on SK Hynix, Samsung, and ; packaging, switches, optical modules, liquid cooling—all backed by dozens of suppliers. The ability to twist such a long supply chain into a tight rope and push it forward generation by generation each year depends on Jensen Huang himself. His vision is also 's most valuable asset. In 2006, he went against the crowd to develop CUDA, was criticized by Wall Street for ten years, and later the whole world’s AI ran on it. NVLink, full rack solutions, from selling chips to selling entire AI factories—every step was first seen and bet on by him. These things don’t show up in financial reports or get counted in DCF models. Many compare and Steve Jobs, saying that after Jobs left, still rose to trillions in valuation. But when Jobs left, Cook had already been groomed for many years, and the supply chain was built by him; the successor was already in place. cannot find a second Cook. Currently, the CFO is temporarily acting as CEO, and the board is searching both internally and externally, but there isn’t a single candidate yet. The interim management can maintain the existing business, but who will decide the future direction? Who will negotiate capacity with and memory manufacturers? Who dares to approve the next CUDA-level big bet? All of this is unknown now. Compared to other MAG7 companies, is the most dependent on its founder. , , and have all changed CEOs, relying on platforms and systems. Much of ’s innovation pace over the past decade has been set by one person. So my concern is not just about alone. The AI revolution narrative in recent years has largely been told by Jensen Huang himself. When he talks on stage about computing power demand and AI factories, the entire semiconductor hardware sector’s valuation rises accordingly. Without its flagbearer, this narrative may suffer a heavy blow, and the short-term impact could be far beyond just one company. Speechless, I didn’t expect the Americans to go so far as to cut off their own arm—truly a divine person.

In this round of heavy spending by the new cloud, the ones who ultimately make money are still the few companies selling the shovels, but the share each gets varies greatly. It's best to do the math before getting on board. plans capital expenditures of $40 to $45 billion next year to build over 1GW of new capacity; is at $37 to $39 billion this year. Just these two companies are spending seven to eight hundred billion annually. Where the money flows can be referenced by two cost breakdowns. One is from Jensen Huang himself in a conference call last August: building a 1GW AI data center costs $50 to $60 billion, with about $35 billion spent on chips and systems. The other is Bernstein's estimate from last year, which is much more conservative, putting the total for 1GW at about $35 billion. A GB200 rack costs $5.9 million, with $3.4 million for compute hardware and $2.5 million for physical infrastructure like the data hall, power, and cooling. Breaking down the project, GPUs account for 39% of total expenses, networking about 13%, memory, wafer foundry, and semiconductor equipment each 3% to 4%, and storage only 1% to 1.5%. The two figures differ by nearly a factor of two, but the direction is consistent: the biggest share goes to . The remaining money is divided into several parts. Within the 13% for networking are switches, optical modules, and cables, which $ANET and benefit from. Physical infrastructure accounts for over 40%; the more the new cloud moves toward self-built campuses, the heavier this part becomes. itself said that from 2027, most new capacity will come from its own campuses. After that, memory and storage come into play. Speaking of storage, I am a firm bull, but one thing must be acknowledged: according to Bernstein's breakdown, storage accounts for a very small portion of the bill in a single data center. I am optimistic about and , relying on tight supply and price increases, which has little to do with their share in the bill. These stocks have high price elasticity; when sentiment rises, they surge, and when it turns, they fall quickly. Look at the past few days. From the close on 11/5 to the close on 11/11, over four trading days, rose 23.6%, 21.2%, 38.1%, and $ANET 23.6%. The next day pre-market, storage and optical modules both pulled back together. This kind of rise is buying sentiment recovery; fundamentals haven't changed that much in a week. So I clearly separate being bullish from chasing highs. I have long been bullish on storage and optical communications as well, but for both, I stick to my set price levels and buy step by step, not chasing when they rebound halfway. The new cloud's heavy spending won't stop in the short term; for the few companies making money, wait for prices to return to your levels before buying more—there's still time.

Jensen Huang has resigned. After the midterm elections, the Democrats' good tactics brought a bunch of scandals, and they just kicked him out, right? At 8:48 AM Eastern Time on 11/12, announced: Jensen Huang resigns immediately from the positions of President, CEO, and Board member. CFO Colette Kress will serve as interim CEO, and the board has started looking for a permanent successor, considering both internal and external candidates. The company also established its first-ever Chairman position, held by Harvey Jones, who was formerly CEO of $SNPS. 33 years, one person worked from start to finish, and now he's gone just like that. Many new investors might not know how came to be. Founded in 1993 in a Denny's restaurant, it was a small company making gaming graphics cards. A few years ago, it almost went bankrupt, with only enough money to last a month. In 2006, Huang defied opposition and invested heavily in CUDA. Wall Street criticized him daily for wasting money and dragging down gross margins, and the stock price fell to single digits. Ten years later, with deep learning rising, everyone realized that the world's AI training tools had long been laid out by him in advance. From a nearly bankrupt graphics card company to a $4 trillion market cap and the engine of the entire AI era, the biggest variable in this story is him. Who will replace him now? Kress is a very capable CFO who has worked with Huang for 13 years, keeps the books clear, and is trusted by Wall Street. But she manages money, not direction. Interim CEO and CFO at the same time, one person doing two jobs—this arrangement itself shows the board has no ready successor. This is exactly what I have been worried about: has never named a number two. When Jobs left in 2011, Cook had been groomed for many years, and only dropped 0.65% the next day. has no Cook. Will he come back? Jobs returned 12 years after being ousted because was struggling. This time, Huang even gave up his board seat, so the chance of a short-term return is very small. Unless the new CEO screws up and the company really hits big trouble, that would be years down the road. Let's talk about the timing. Data center approvals, AI regulation, House hearings, independent investigations—all these events clustered in the two weeks after the election, resulting in the AI flag bearer stepping down. Many suspect a setup behind this; I also think the timing is too coincidental. Until the investigation concludes, this remains suspicion without evidence, but the political climate is clearly unfriendly to AI. So what about ? The company won't collapse just because the CEO changed. Customer orders remain, capital expenditures from , , and continue to rise, and Blackwell and Rubin's roadmap has long been planned. The earnings report on 11/17 will proceed as usual. What’s hurt is the valuation. Ives calculated before that with Huang, it’s worth 300; without him, 225. The difference is the "key person" factor. The market will cut first and then gradually assess the new management's quality. The AI revolution won't stop because of one person, but the short term will definitely be more volatile. My approach: no leverage, no short-term options betting on earnings, no panic selling during the worst emotional days, and no rush to bottom-fish. My first purchase will only be considered near the price I set before, waiting for the 11/17 earnings to see how Kress and management talk about the future, and who the new CEO is, then proceed step by step. The greatest company has changed its captain; the next few months will test how strong its system really is.

This drop in has laid all the problems of Xinyun stock on the table: no matter how good the orders are, the stock price still follows the financing sentiment. On 11/11, it closed up 15.4%, but the next day after the earnings report came out, it fell back a large chunk pre-market, fluctuating about 30% back and forth. The report itself is quite impressive. Q3 revenue was $981 million, up 571% year-over-year, adjusted EBITDA was $466 million, both exceeding expectations by a good margin. What made the market turn was the guidance. This year’s revenue can only reach the low end of the $3.0 to $3.4 billion range; analysts originally estimated $3.35 billion. Capital expenditure was raised to $27 billion this year, previously given as $20 to $25 billion; next year it’s directly set at $40 to $45 billion, while analysts only estimated less than $29 billion. Next year’s revenue is expected to be $12 to $13 billion, and pre-tax breakeven is not expected until the end of 2027. The company explained that 75% to 85% of next year’s spending can be covered by customer prepayments, asset-backed financing, and cash on hand; the rest, the market will have to figure out whether it’s debt or equity issuance. Looking at the four companies together, the differences are actually quite large. has the thickest order backlog, $132.5 billion, and this year’s revenue guidance was even raised to $13.0 to $13.3 billion. But it carries about $47 billion in debt, with Q3 net interest of $846 million, and Q4 interest guidance of $1.05 to $1.15 billion. Strong orders, but also the heaviest leverage. has remaining performance obligations of $38.7 billion, much thinner than , with $7.4 billion cash on hand, and borrowed another $5.75 billion in Q3. Its trouble lies in the pace, spending money faster than signing orders. has a backlog of $20.3 billion, the company says about two-thirds are investment-grade clients. It’s stuck on delivery: the capacity for the batch was delayed due to customer network design changes and supplier delays, pushing the $4 billion annualized revenue target from the end of this year to January next year, possibly February. is more like a landlord. 1.41 GW of leases, about $36 billion contract revenue in the initial lease term, starting at 15 years, tenants are and investment-grade mega clients. Debt is $6.4 billion, with the latest guaranteed notes at 7% interest. Ranking by order firmness, I’d put and first, ’s orders are not bad but delivery is delayed, and has the greatest spending pressure. But on 11/11, all four rose together: up 24.7%, up 7.4%, up 5.6%, then all fell together pre-market the next day. No one distinguishes whose orders are firm in the short term; everyone is trading the same thing: whether the money can still be borrowed, and whether it’s expensive. My approach to these stocks is simple: small positions, buy in several times at pre-set price levels, don’t chase on days they rise 15%, don’t panic sell on days they fall 15%, and definitely no leverage. Rather than focusing on whether quarterly revenue beats expectations, I pay more attention to where their bond yields and financing costs are headed, whether valuations can hold steady, and ultimately that’s what matters.