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

769followers

Feed

Livestream

Feed

Without Jensen Huang, will continue to make money and grow, but it most likely will never return to the greatness it had under Jensen Huang. First, let's talk about why it can't be replaced. In this AI wave, almost everyone is making money around . The combined capex guidance of , , , and this year is between $720 billion and $740 billion, with the largest portion spent on GPUs and full server racks. 's CoWoS advanced packaging, $MS estimates that by 2026 about 60% of global demand will be for . On the HBM side, $UBS estimates nearly 60% of global demand this year comes from ; SK Hynix, , and Samsung are racing to expand production, with the largest order being from . Further down the line, and make network and optical modules, and assemble servers, and new cloud providers like and rent computing power with its GPUs—all hanging on this supply chain. Its own financials are solid. As of the quarter ending July, revenue was $96.2 billion, more than doubling year-over-year, with data center revenue at $89 billion accounting for over 90%, and a gross margin of 75%. The company’s guidance for the quarter to be announced on 11/17 is $108 billion, and it says next fiscal year revenue could grow about 70%, even with supply constraints. CUDA and its supporting software have over 7.5 million developers. These assets won’t disappear just because of a CEO change tomorrow. Orders are there, industry dominance remains, and I have no doubt about these two points. But ’s growth from hundreds of billions to trillions in three years relied on Jensen Huang. CUDA was his decision to bet on, and those years it was often criticized as burning cash. Later, he bet on AI training, shifted to selling full server racks, set a yearly generation cadence, and pulled the supply chain and new cloud providers together. Specifications for the next generation, how many layers of HBM, how much packaging capacity—he basically saw and bet on these first, and the entire chain expanded production following him. A big part of ’s market valuation is buying into this person’s vision. Now sitting there is interim CEO and CFO Colette Kress, with newly appointed Chairman Harvey Jones. She can maintain the numbers for these quarters, but where the company goes in the next ten years and who will make those decisions—no one knows. There’s a heated debate on X today, roughly in several camps: Some say AI demand hasn’t quit, CUDA remains, and the drop is a buying opportunity; Some compare it to after Steve Jobs left, saying it will just be a cash machine with no major innovation; Some say the performance is fine, but the problem is how much valuation the market will still give; Some think customers will accelerate self-developed ASICs and turn more to , causing market share to be lost faster; Others directly compare it to in 2018, saying the board made the biggest mistake of their lives. I basically agree with the first three: orders remain, moat remains, valuation will be discounted—these three can coexist. The last two are exaggerated; market share will be slowly lost, but ’s current capacity, ecosystem, and customer stickiness are completely different from back then. In the short term, I’m actually optimistic. The bad news seems fully out today; the stock dropped to 148.80 at one point at open, then recovered to around 164 in the afternoon, sentiment is repairing. The 11/17 earnings report has orders already lined up and is likely to be strong, so the stock is more likely to go up. Long term, I really can’t be sure. At times like this, I think of Jack Ma and $BABA. Jack Ma stepped down as chairman in September 2019, and $BABA still rose for more than a year after. On October 24, 2020, after his speech at the Bund, $BABA closed at 317.14 three days later, its all-time high close. On November 3, Ant Group’s IPO was halted, and Jack Ma basically faded away. $BABA then fell from that high to 63.15 by October 24, 2022, losing 80% in two years. The company kept making money during those two years. Revenue grew 41% in the fiscal year ending March 2021, 19% the next year, and only 2% the year after. The business didn’t collapse, growth slowed year by year, new business was rolled out more cautiously, and the market stopped paying for its future. Today, $BABA is just over 90. Both cases have political shadows. $BABA’s is obvious; for this time, I think it’s hard to say politics is unrelated, but that’s just my own judgment. The differences are also clear. stands at the peak of industry demand, customers plan to increase capex next year; it’s in the U.S. market with much more transparent rules; CUDA plus locked-in capacity make its moat much stronger than e-commerce. So I don’t think will follow $BABA’s path of an 80% drop. But "the founder leaves, the company still makes money, but the market no longer pays a premium for the future" will likely partially repeat with . Operationally, I’ll stick to the usual: no leverage, no short-term options bets on 11/17, keep cash, and buy in batches at my price. will probably remain the most profitable AI company, but the kind of greatness under Jensen Huang’s leadership will be hard to replicate.

Jensen Huang's situation is definitely not a small matter, but to say it's just because of some personal issues, I don't believe it. There are always a few personal scandals on Wall Street every year. When a CEO has personal life problems, internal investigations, apologies, and quiet handling usually follow. Most companies go through this process. But for , on 11/8 the board still expressed confidence in him, and by 8:48 AM ET on 11/12, the announcement came out: the president, CEO, and all directors stepped down; CFO Colette Kress temporarily became CEO; former $SNPS CEO Harvey Jones became the company's first independent chairman. Within four days, the 33-year founder didn't even keep a seat on the board. This speed makes it hard not to think politically. The midterm elections just ended, and several congressional committees had already summoned him for a closed-door briefing on 11/18, with lawmakers repeatedly mentioning his name. I think he is a scapegoat pushed out in this round of political struggles; the personal issues are just the knife handed over. There are all kinds of rumors flying around on X now, roughly these main ones: Some say the Democrats started a purge after the election, targeting tech tycoons close to Trump, with Huang being the first. Some say it's a scheme by short sellers and Wall Street to first crash the stock price to accumulate shares, then bring him back in a few months to push the price up again. Some English speakers mention the deep state, saying it's aimed at open-source AI. Others say he voluntarily "escaped like a cicada shedding its shell." These are all guesses without any evidence; the last two sound like jokes to me. I believe the political angle myself, but that's just my judgment. No one can be sure before the investigation concludes. without Jensen Huang is a soulless . is still a good company; the CUDA ecosystem, existing orders, and capacity locked with won't disappear just because of a CEO change tomorrow. But the premium the market has given it over the years was for a person who could see the next step two or three years ahead of everyone else and dared to bet heavily. Vision, innovation, leadership—these prospects are now unclear. The interim CEO can't provide them, and it's hard for an outside professional manager to do so. So my judgment is straightforward: will most likely decline from here. It's very difficult to return to the new high of 239 on 10/6. Even if it does, the path will be very, very tortuous. What's more troublesome is the string of suppliers behind it. In recent years, hasn't just been selling chips; it is the overall coordinator of the entire AI hardware chain. The next-generation architecture specs, how many layers of HBM, how much CoWoS capacity, how to set up racks, which route the optical modules take—basically, Jensen Huang decides, and the whole chain expands production following his roadmap. 's advanced packaging expansion has as its top client. On the storage side, , SK Hynix, and Samsung are desperately expanding HBM, with as the biggest buyer, following Rubin's rhythm. For optical modules and networking, , , , etc., the schedule for speed upgrades largely follows 's rack plans. Further down are , , and others assembling servers. Orders won't disappear overnight. The reason everyone dared to invest heavily in expansion years in advance was trust in that person's roadmap. Now that the helmsman is gone, whether the roadmap will change, the pace slow down, or customers hesitate—no one can guarantee. Once certainty is gone, the first to get hit are the suppliers' valuations, because their premiums were based on 's roadmap. If declines, this chain will likely suffer even more. Then there's the 11/17 earnings report, which is really difficult now. I'm not too worried about the numbers; orders and capacity have long been scheduled, so this quarter probably won't be bad. The difficulty lies in the earnings call. Previously, Jensen Huang alone could weave next-gen architecture, customer needs, and sovereign AI into a two- or three-year story that the market bought into. This time, the interim CEO and CFO will be there. She can explain the numbers clearly but can't provide foresight. If guidance is too high, the market won't believe it; if conservative, it will be interpreted as "without Huang, it can't work." Options have already priced in this uncertainty. The straddle implied volatility for 11/20 is about ±10%, IV in the 80s or 90s. Buying short-term options to bet on earnings now is just giving premiums to sellers; even if the direction is right, IV crush could wipe out gains. I won't touch it, won't leverage, will buy the stock in batches at price levels, keep cash, and use any rebound as an opportunity to reduce holdings. Finally, will Huang come back? Short term, basically impossible. The president, CEO, and directors all stepped down; the board set up an independent chairman for the first time—this is a cut-off. Independent investigations, shareholder lawsuits, and Congress matters are still unresolved; no one dares to bring him back now. If he really returns, there are only a few paths: The investigation clears his name, the new CEO messes up, and major shareholders and the board pressure to bring him back. Like Jobs, who left in 1985 and returned in 1997 after 12 years, when was almost failing. Alternatively, he returns as a founder or advisor, with influence but no direct control. Or the board is completely reshuffled, bringing in a new group. All these have low probabilities. On Polymarket, even "he starts another company before year-end" has 30% backing, but no one is betting on his return to . The market's attitude is clear. So I see this as a political event. is a good company without a soul now, and it and its supply chain will have to be repriced going forward.

Will Jensen Huang come back? My answer is: basically impossible in the short term. He left too decisively this time. He handed over the CEO, president, and board seats all at once, and the company even specially appointed the first-ever independent chairman, Harvey Jones, who was formerly the CEO of $SNPS. This board operation is meant to completely cut ties with him, not to have him step aside temporarily. Moreover, the investigation commissioned by the board from a law firm has not concluded yet, and the House hearing and shareholder lawsuits are still pending. Before these are resolved, anyone who dares to bring him back is just inviting trouble for themselves. Some might cite Steve Jobs as an example. He was ousted from in 1985 and returned in 1997. But that was a 12-year gap, and was on the brink of collapse before they had to bring him back. currently has full orders and no "must-have him" pressure. If Jensen Huang were to return, it would probably only be under one circumstance: the new CEO screws up, the company really runs into big trouble, and the investigation clears him. That would be years from now. So who will save without Jensen Huang? For now, the interim leader is CFO Colette Kress, who has worked with Jensen Huang for 13 years and is highly trusted on Wall Street. But she manages the money, not the direction, and the market generally sees her as a transitional figure. The internal favorite is Ian Buck. He helped develop CUDA back in the day, now manages data center and supercomputing business, and has been with for nearly 20 years. As of this morning, Polymarket gives him a 34% chance, Kalshi 42%, both ranking him first. The problem is, Buck is a top engineer, but Jensen Huang does more than just technology. Negotiating capacity with , dealing with governments worldwide, telling the AI story on stage to the world—Buck has never done these. Others mentioned include Jay Puri, who manages global sales, and co-founder Chris Malachowsky. External names include CEO Rene Haas, and even ’s Lisa Su. But Lisa Su is doing well at , and the more chaotic is, the more advantage she gains, so no reason for her to jump ship. The most interesting is Tim Cook. At one point, Kalshi had people betting on Cook to take over , with odds reaching 91%, then dropping back to about 10%. He is currently chairman of and indeed the world’s top at managing global supply chains. But can Cook come? I think it’s very difficult. First, he may not want to. Being chairman at and having achieved success, there’s no need to jump into a mess still under investigation and hearings. Second, even if he came, needs more than just someone who can manage supply chains. When Cook took over , the product direction was set by Jobs, and the iPhone was already ready. What needs now is exactly someone who sets direction and dares to make bold bets. Ultimately, Jensen Huang carries three roles: chief engineer, chief salesperson, and chief storyteller. Whoever comes from outside can at most fill one of these roles. So the most likely outcome for is to switch to a team to carry the load: a technical person like Buck managing products, Kress and Puri managing money and customers, and the chairman managing governance. The business can be maintained, but the speed of one person making decisions and the entire industry chain following will probably not return. For the stock price, this means a period of uncertainty. Before the new CEO is chosen, the valuation will be hard to truly recover. My approach: no leverage, no betting on earnings, buy in batches according to price. On 11/17, see the earnings report and what management says, who the new CEO is, then reprice it.

There is something very strange about today's session: was slammed down 14% at the open, while the broader market barely moved. Usually, when drops this much, the Nasdaq would at least fall by two or three points. Today, QQQ closed down only 0.32%, and SPY down 0.34%. The largest weighted stock was hammered down, yet the index just wouldn't fall. This is exactly the feeling I describe as the bad news being fully priced in. You can also see it on the candlestick chart. QQQ's intraday low was 679.84, breaking below the 200-day moving average at 682, then it pulled back all the way to close at 688.17, leaving a very long lower shadow. SPY touched 753.70 and closed at 757.25. The semiconductor index opened lower at 10174 and closed at 10506, rallying more than three points from the open. The selling pressure was absorbed. Let's rewind to November 5. That day, SPY closed at 733, QQQ at 647, and the semiconductor index at 9011 — the most panicked day of this cycle. In the following week, one piece of bad news after another came out — Jensen Huang's issues, political matters, and so on — but every time the index dipped, it never touched the November 5 levels. Now SPY is 3.3% higher than that day, QQQ 6.4% higher, and the semiconductor index 16.6% higher. Regarding oversold indicators, I have to be honest. The indexes are no longer oversold. SPY's RSI is 48.6, QQQ 45.4, and the semiconductor index 45.6 — all in the middle range. The real oversold condition was around November 5. Today, the only seriously oversold stock is , with an RSI of 26, closing at 166.19, just touching the lower Bollinger Band at 164.75. So what the market did today was to retest the November 5 bottom to see if it holds. The options market gave the answer even earlier than the candlestick chart. On November 5, the 9-day VIX surged to 32.66, higher than the VIX at 28.18, indicating short-term panic outweighing long-term — a classic panic market. Today, the 9-day VIX is 17.84, VIX 18.97, and the 3-month VIX 20.23, returning to a normal ascending order from low to high. The VIX dropped from 28 to 19, and the options market no longer expects a major event in the coming days. Tomorrow is PPI, and the options-implied volatility is not high. The straddle for SPY expiring on November 13 implies about ±0.75%, roughly 752 to 763; QQQ about ±1.3%, 679 to 697. The largest put open interest for QQQ this week is clustered around 650, almost exactly the November 5 low of 647. Big money has placed protection there, not chasing lower. So tomorrow, the key is how the market reacts to the data; whether PPI is hot or cold is secondary. The October PPI will be released at 8:30 AM ET, with expectations of 0.4% month-over-month and 0.3% core. Even if PPI is a bit hot, as long as QQQ holds today's low of 679.84 and SPY holds 753.70, and the market can recover intraday after a dip, the phase bottom from November 5 to today is basically confirmed. If the data causes QQQ to break below the 200-day moving average at 682 and it can't close back above, then the market will have to retest the November 5 lows around QQQ 647 and SPY 733. Even if the bottom holds, I only consider it the bottom of a trading range. The political issues are unresolved, 's earnings on November 17 haven't come out, and there is still the November 9 gap and the 50-day moving average overhead; QQQ's 50-day moving average is at 718. Being able to stop the decline is far from being able to rally all the way back. My approach remains the same: no leverage, no betting on PPI with options expiring tomorrow, keep cash on hand. Buy in batches if support holds, wait for the next level if it breaks. How the market moves in the first half hour after tomorrow's open will be more accurate than anyone's prediction tonight.