美股投资young
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
760followers
Feed
Livestream
Feed
I really couldn't hold it in
🤣🤣🤣
$NVDA
To be honest, as long as they don't divorce and it doesn't affect company management,
it's not a big deal, just a moral issue.
I don't think it's espionage, but smuggling might be possible 🤔
Today, the Philadelphia Semiconductor Index rose 6.7%, but this is a rebound, still far from a reversal.
First, let's talk about how to distinguish them. A rebound happens when the price has fallen too much, shorts cover, and bottom-fishing capital enters, causing the price to bounce back somewhat; a reversal means the reason for the decline is gone, and capital is willing to hold long-term positions again. To judge which it is, don’t just look at how much it rose in one day; look at three things: whether the reason for the decline still exists, whether the previously broken support level can be recovered, and whether the price can hold during a pullback.
Let's check these three points against the current situation.
First, the reason still exists. This round of semiconductor sell-off is mainly the market pricing in the possibility that after the Democrats took office, data center approvals slow down and AI regulations tighten. Up to today, neither side has given any statement—no confirmation that it’s just a delay, nor any exclusion of continued tightening. The cause of the decline remains; what’s risen is just sentiment.
Second, technically it has only reached the first hurdle. $SOXX gapped down on November 5; the previous day’s low was 147.23, but the high that day was only 143.73, leaving a gap. Today’s intraday high was 146.98, just a few cents short of filling the gap. The Philadelphia Semiconductor Index is similar: the low on 11/4 was 10064.73, today’s high was 10059.68, just a bit short.
Filling the gap is the most common first target of a rebound, and usually there will be selling pressure there. Above that, there are two more hurdles: one is the July 29 low at 10447, which was support before the break but now acts as resistance; the other is the 20-day moving average, around 10900. Above that, the gap left on October 30 between 10957 and 11364 has not yet been touched.
Third, there has not been a pullback yet. After a reversal, there is usually a pullback that does not break the previous low, followed by a higher high. The Philadelphia Semiconductor Index is still just over 30% below its highest close, having just bounced from the low; this step has not happened yet.
$NVDA also illustrates a point today: while the Philadelphia Semiconductor Index rose 6.7%, it fell nearly 5% due to a scandal. Capital is currently selective in buying, not fully returning.
So I treat today as a normal oversold rebound, not a reversal signal. What would really make me change my view is a clear political statement, plus the price standing back above 10447 and the 20-day moving average, and a pullback that does not break support. If one of these happens, I will raise my confidence; if all happen, then I would call it a reversal.
The biggest mistake at times like this is to treat a big one-day rise as a confirmed bottom and then leverage up to chase. Rebounds can make money fast, but pullbacks come fast too. Even with leaders, buy in batches, don’t use leverage, and wait for both the gap fill and political clarity to give the answer.
The panic over halted construction is loud, but it did not appear in the power company's guidance. $CEG released its Q3 earnings today, with adjusted profits exceeding market expectations and an upward revision of the full-year profit forecast, causing the stock price to rise about 3% intraday.
This company is the largest nuclear power operator in the United States and has signed many long-term power supply agreements with major companies like Google and Amazon recently. If, as the market worried last week, all data centers nationwide were to halt construction, the first to revise their outlook should be companies directly supplying power to data centers. Instead, it not only did not downgrade but actually raised its expectations.
This, of course, does not mean political risks have disappeared. Resistance to data center power usage exists in various states, and approval processes may slow down. But power companies sign long-term contracts and consider electricity demand for many years ahead; their judgment is more telling than a single day's stock price.
AI demand remains, and politics is more about changing the pace and cost of construction, not yet to the extent of directly cutting demand. The power company has provided an answer with its own guidance, which is more convincing than slogans.
CoreWeave's earnings report is tomorrow. I'm more concerned about how it will raise funds for upcoming expansion; this quarter's revenue is secondary. AI demand remains unchanged; however, companies relying on borrowing to build data centers will feel political and interest rate pressures earlier than others.
$CRWV's current situation is typical. Its stock price has dropped about half in the past month, similar to Nebius, which also expands through debt. CNBC reported that yields on some of its bonds maturing in 2031 have exceeded 16%, and the five-year credit default swap spread rose about 50% in a month. The bond market prices are more straightforward than stock prices: lenders now demand higher returns to continue lending.
The company is, of course, firefighting. The CEO publicly stated that no political changes are expected to affect existing customer commitments and backlog orders. It also wrote to the incoming Texas governor, expressing willingness to cooperate, responsibly complete permitted projects, and establish clear, long-term standards for future development. This approach is smart: it proactively positions itself as part of the solution without confronting politics head-on.
But the earnings report must answer questions more complex than just "orders are unaffected." As of the end of June, the company had about $5.5 billion in cash and approximately $35 billion in debt, including a $1.3 billion loan with a 15% interest rate. Analysts generally believe it won't face a short-term payment crisis and has enough cash to fulfill existing contracts; the issue is that growth in the coming years requires more capital. High interest rates, lenders becoming hesitant, and increased uncertainty in data center approvals—all three pressures weigh on a company that needs continuous financing, revealing the cost.
So tomorrow, I will focus on three things: whether new financing costs and sources are secured, if capital expenditure plans have been adjusted, and whether the pace of converting backlog orders into revenue has changed. The revenue figure itself, with market expectations around $3.55 billion, is only part of the story; if the financing path is unclear, even strong revenue won't reassure the market.
This is why I've never liked touching small-cap compute power stocks with high leverage and financing-driven expansion in this environment. Suppliers like $NVDA get paid immediately after selling chips, and large cloud providers have their own cash flow; CoreWeave, however, must borrow money to buy equipment first, then wait for customers to pay rent slowly. Any bottleneck in this chain puts pressure on it. AI demand is not the problem; the challenge is who bears the funding risk during the build-out phase.
I won't bet on its earnings result. Its experience illustrates how tight the entire industry's funding chain is. If even companies like this can smoothly find money and maintain expansion pace, that's good news for the whole AI build-out; if raising funds becomes increasingly difficult, the hardest hit will be these companies, not the leaders.
The performance of $COHR this week explains better than any research report just how difficult it is to capture the current state of optical communications. Or perhaps the optical module landscape has changed.
On the day the earnings report came out, it closed down about 13%. The numbers themselves are actually quite impressive: quarterly revenue of $2.37 billion, a 50% year-over-year increase; data center and communications business grew 79%, already accounting for 82% of total revenue; next quarter revenue guidance is $2.55 billion to $2.75 billion, with full-year growth expected to exceed 60%. Management also said the first $3 billion revenue quarter will appear within this fiscal year. The CEO's exact words were that AI data centers are shifting from copper connections to optical connections, and Coherent's platform and manufacturing scale can help customers meet bandwidth and energy efficiency demands.
So why did the stock still fall? Because the cash didn't keep up with the profits. Operating cash flow was negative this quarter, with capital expenditures and inventory both increasing. The market sees that growth requires upfront investment and stocking, and Bank of America even cut the target price to $240.
Then it rebounded about 14% the next day, and by 11 a.m. Eastern today it rose nearly 18%, with an intraday low around $197 and a high near $228. Citi and Stifel both raised their target prices.
This is the current rhythm of optical communications: demand is certain, but the stock price changes daily. After Marvell announced a stock-for-stock acquisition of Lumentum, $MRVL and Lumentum's stock prices have been linked, both rising about 11% and 13% today respectively. The only independent large optical module companies left are Coherent and a few others, so the market is naturally more sensitive in pricing them.
I am optimistic about optical communications in the long term; as computing power scales up, the data between chips and cabinets will increasingly rely on optics. But for a stock like Coherent that falls 13% in a few days, then rises 14%, and today goes up 18%, I would rather accumulate slowly in batches, adding more when cash flow catches up with profits, rather than chasing heavily just because it rose 18% in one day.
OKB has dropped more than 10% in the past 7 days, HYPE has fallen nearly 12%, and XRP has also continued to weaken. BTC is temporarily holding above $75,000, but these coins have not shown synchronized recovery, and capital support remains insufficient.
$OKB is currently around $102, with $100–$101 as the first support level and resistance at $104–$106. Only by firmly reclaiming $106 can the structure improve; breaking below $100 means a risk of further testing $97. OKB still has a significant gap from its previous trading range above $110.
$HYPE is currently around $77.7, with $76–$77 as initial support and resistance at $79–$80. Only by firmly reclaiming $82 can the recovery expand; losing $76 may lead to further testing of $74. HYPE’s previous strength has clearly weakened, with rebounds continuously facing selling pressure.
$XRP is currently around $1.16, with $1.15 as the first support and resistance at $1.18–$1.20. Breaking above $1.20 is needed to improve the short-term structure; breaking below $1.14 risks further decline. Compared to OKB and HYPE, XRP’s recent volatility is relatively mild but similarly lacks active buying pressure.
ZEC has risen about 4% in the past 7 days, while DASH is still declining, and NIGHT's weekly chart has also not turned positive. The related tokens have not formed a unified upward trend, with short-term funds more concentrated in ZEC.
$ZEC is currently around $1275, down more than 1% in 24 hours. The $1238–$1250 range is the first support zone, with resistance at $1300 above. Only breaking through $1300 will there be a chance to challenge $1350; falling below $1230 requires caution for profit-taking to continue. ZEC's weekly chart still leads, but short-term volatility is obvious.
$DASH is currently around $47.2, with a slight increase in 24 hours. Support is first seen at $46–$46.5, with resistance at $48–$50 above. Only by firmly holding above $50 can the structure improve; breaking below $45.7 may lead to further testing of $44.
$NIGHT is currently around $0.0405, up about 2% in 24 hours but still down about 3% over the past 7 days. $0.039 is the first support, with resistance at $0.041–$0.043 above. Only breaking through $0.043 can the recovery expand; falling below $0.0383 requires caution for further decline.
On the day Nvidia ran into trouble, Google released two new TPUs, and the market reaction was very direct: Nvidia dropped more than 5% intraday, while Google rose about 2% to 3% intraday. Many people saw this contrast and their first reaction was that Google was taking advantage of the chaos to grab business. But this trend didn’t just start in the past few days; TPU has been accelerating all year, it’s just that today it was put under the spotlight.
First, what was released today. $GOOGL launched the TPU 8tx and 8i Max, one designed for regulated customers like governments and banks that require encrypted computing, paired with AMD processors; the other aimed at tasks needing to handle ultra-long contexts. Both are reconfigurations of existing chips, shipping next year. Along with the 8t and 8i released in April this year, this is the first time Google has launched four TPU systems in one year. The product line is becoming more segmented, indicating it’s starting to make chips tailored to different customer needs, which means it’s seriously doing business.
A bigger change is in the sales approach. Google only started selling TPUs to external customers in Q2 this year, and in October announced that Oracle will offer TPUs on its own cloud starting in 2027, making Oracle the first major cloud provider outside Google to do so. On the customer side, Anthropic officially announced it will use up to one million TPUs, and OpenAI is also reported to plan direct purchases. TPU has evolved from a chip used only by Google itself to a supply channel others can buy from.
So I prefer to interpret today as a TPU repricing. In the past, the market mainly looked at Google for search, advertising, and cloud; now there’s an emerging chip business. Customers want to find more options, which is normal: computing power is too expensive and tight, and no one wants to put all their demand on a single supplier. Anthropic itself uses TPU, Amazon’s Trainium, and Nvidia GPUs simultaneously, and this combination itself tells the story.
But I wouldn’t write this as TPU replacing $NVDA. Oracle is also expanding Nvidia system deployments at the same time; TPU is a parallel choice and cannot replace GPUs. Nvidia’s software ecosystem, complete system, and customer base remain intact. Today’s drop mainly comes from uncertainty caused by the board investigation, unrelated to product obsolescence. Customers have one more choice, and Nvidia’s bargaining power might be slightly squeezed—that’s the real impact; the demand itself hasn’t been taken away by anyone.
$AVGO is also involved here; it’s Google TPU’s design partner. The more TPUs sell, the more opportunities Broadcom’s custom chips and networking business have. This chain benefits both Google and Broadcom together, not just Google alone.
AI computing demand is still there, and it’s so large that one company can’t supply it all, which is why TPU external sales and Oracle reselling happen. To me, today’s contrast perfectly shows this track is big enough to accommodate more than one winner. I still lean towards the leading players making phased deployments and won’t chase Google or cut Nvidia just because of one day’s price moves. Which ones to track specifically and in what order will continue to be updated in the members’ group.
