$AMD earnings report: Management expects CPU revenue to still grow by more than 70% in 2027, continuing growth on a higher base. AI GPUs can rapidly scale revenue, but new product ramp-up, HBM, and whole system costs will all impact gross margin. EPYC is relatively mature, with a more diversified customer base and more stable profitability. EPYC is AMD's CPU brand targeting servers and data centers, mainly competing against Intel's Xeon. The fundamentals of the CPU segment are solid; the issue still lies with long-term interest rates. $INTC $ARM

Three incense sticks above the head, if not dead, at least injured $PLTR is out first Next, either break through and retest, or honestly wait to buy in the range below 130 #财报观察员：业绩喜忧参半，解禁将至！SpaceX后续怎么看？ #SpaceX首份财报超预期，解禁仍是关键变量

$SKHY Short-term institutions are buying insurance, while long-term funds have not given up on bullishness. 1/ Near-term defense is very heavy. On August 7, Put OI was nearly 4.8 times that of Call, and about 3.4 times on August 21. Institutions are clearly guarding against high volatility and rebound failures in the coming weeks. 2/ But the long-term structure is actually more bullish. Call OI for September, December, and January 2027 is clearly dominant. Especially in January 2027, Calls increased by about 10,699 contracts, while Puts only increased by about 248 contracts. This does not look like an outright bearish stance, but rather holding long-term longs while buying Puts to protect against short-term risks. 3/ In bulk, institutions bought October $135 Puts while selling $180 Calls. This protects against risk below $135 while still willing to hold near $180. 4/ The biggest short-term issue is negative Gamma. $141 is the Put Wall, $145 is the Dealer acceleration point, and $160 is the Call Wall. Breaking below $141 may accelerate a move toward $135. Holding steady between $145–$150 is needed for a chance to repair toward $160. The long-term logic remains, but short-term insurance is heavy. Before $150, it is still a rebound test; $160 is the next real resistance.

Currently, the US stock market is basically in an Amplifying + negative Gamma volatility amplification state, coupled with long-duration interest rates at high levels, so sharp rises and falls won't be too surprising. Next Wednesday is the US Treasury auction, which could be even more explosive.

$MU options data actually isn't much to look at, mainly for hedging. There is a possibility of a short-term oversold rebound, but until it retakes $800, the rebound still lacks structural confirmation. A bigger issue comes from the huge negative Gamma expiring on July 31. Dealers will amplify price volatility, making chasing gains prone to quick pullbacks, and bottom-fishing may also face a secondary sell-off. In terms of options structure, near- and mid-term open interest is still mainly defensive. Long-term bulk Calls remain bullish, and the long-term logic has not been completely broken. Spot strategy is better suited to wait until $800 is firmly retaken before following, don’t catch a falling knife. $800 is the rebound confirmation level $750 is an important defense level Before the U.S. Treasury auction on August 5, it’s actually a good time to rest.

KLA $KLAC latest earnings report, good news: they have already started signing capacity agreements for 2029. Bad news: it's still duration income. FY2026 Q4 revenue was $3.658 billion, up 15% year-over-year, Non-GAAP gross margin remained at 62.4%, net profit was $1.39 billion. Service revenue in Q4 was $820 million, up 17% YoY, accounting for 22%, of which 80% are contracts, a rare annuity stream in the equipment industry. Management's long-term service revenue growth target is 13%–15%. ———— In the past 12 months, the company generated $3.77 billion in free cash flow, committing to return over 90% to shareholders, along with a $9.7 billion buyback authorization. The average maturity of long-term debt is about 18.1 years, with an average interest rate of 4.67%, so there is no need to worry about refinancing costs suddenly rising for a long time. The cost of capital is locked in; next, the focus is on the numerator side. Management emphasizes that process control involves many varieties, with small volumes per project. Different fabs, different processes, different defects all require separate debugging and testing solutions. Besides algorithm accumulation, KLA has over 1,600 application engineers stationed long-term at customer sites to help adjust equipment and analyze data. This competitiveness relies on years of experience, software databases, and a large engineering team; even if newcomers build hardware, it is difficult to quickly match the full set of service capabilities. In contrast, lithography equipment is closer to large-scale shipments around the same EUV platform, with a higher degree of product standardization. When KLA says "scale is about 6 times closer to competitors," it mainly refers to core markets like wafer inspection and mask inspection. Although $AMAT and $LRCX often emphasize expanding their process control business, they are still far from KLA in product coverage, installed base, data accumulation, and customer support capabilities. Not yet enough to shake KLA's market position. —————— More notably, the CFO said on the call: "we are sizing the company to be able to serve the more bullish scenarios." We are configuring capacity according to the most aggressive scenarios. Management's discussion of the 2027 WFE baseline has already reached $190 billion, while preparing for a larger industry scale. Some key components with delivery cycles as long as 18–24 months are already negotiating capacity agreements for 2029. Semiconductor equipment management has always been conservative. Before the peaks in 2018 and 2022 cycles, they both stepped on the brakes early. Now openly discussing "bullish scenarios" indicates that customer planning, order visibility, and supply chain signals are strong enough. ———— Of course, this is also a duration commitment. Among the big four semiconductor equipment companies, $KLAC has the longest duration. RPO reaches $12.5 billion, covering about 3.4 quarters of revenue. The 2027 equipment boom and 2029 capacity agreements place more value in the long term. KLA's business quality is increasingly close to that of a software company, making its valuation more sensitive to long-term interest rates. Long-duration assets are sensitive to interest rates. KLA's long-term debt weighted maturity is 18.1 years, interest rate 4.67%, locked in. Refinancing risk is near zero; the rest is all about the numerator story. It's just that with duration rates currently high, $KLAC is having a tough time.

KLA $KLAC latest earnings report, good news: they have already started signing capacity agreements for 2029. Bad news: it's another duration income. FY2026 Q4 revenue was $3.658 billion, up 15% year-over-year, Non-GAAP gross margin remained at 62.4%, net profit was $1.39 billion. Service revenue in Q4 was $820 million, up 17% YoY, accounting for 22%, of which 80% are contracts, a rare annuity stream in the equipment industry. Management's long-term service revenue growth target is 13%–15%. ———— In the past 12 months, the company generated $3.77 billion in free cash flow, committed to returning over 90% to shareholders, and still has $9.7 billion in buyback authorization. The average maturity of long-term debt is about 18.1 years, with an average interest rate of 4.67%, so there is no need to worry about refinancing costs suddenly rising for a long time. The cost of capital is locked in; next, the focus is on the numerator side. Management emphasizes that process control involves many varieties, with small volumes per project. Different fabs, different processes, and different defects all require separate debugging and testing solutions. Besides algorithm accumulation, KLA has over 1,600 application engineers stationed long-term at customer sites to help adjust equipment and analyze data. This competitiveness relies on years of experience, software databases, and a large engineering team. Even if newcomers build hardware, it is difficult to quickly match the full set of service capabilities. In contrast, lithography equipment is closer to large-scale shipments around the same EUV platform, with a higher degree of product standardization. When KLA says "scale is about 6 times closer to competitors," it mainly refers to core markets like wafer inspection and mask inspection. Although $AMAT and $LRCX often emphasize expanding their process control business, they are still far from KLA in product coverage, installed base, data accumulation, and customer support capabilities. Not yet enough to shake KLA's market position. —————— More notably, the CFO said on the call: "we are sizing the company to be able to serve the more bullish scenarios." We are configuring capacity according to the most aggressive scenarios. Management's discussion of the 2027 WFE benchmark has already reached $190 billion, and they are preparing for an even larger industry scale. Some key components with delivery cycles as long as 18–24 months are already negotiating capacity agreements for 2029. Semiconductor equipment management has always been conservative. Before the peaks in 2018 and 2022 cycles, they both stepped on the brakes early. Now openly discussing "bullish scenarios" indicates that customer planning, order visibility, and supply chain signals are strong enough. ———— Of course, this is also a duration commitment. Among the big four semiconductor equipment companies, $KLAC has the longest duration. RPO reached $12.5 billion, covering about 3.4 quarters of revenue. The 2027 equipment boom and 2029 capacity agreements place more value in the long term. KLA's business quality is increasingly close to a software company, so its valuation is more sensitive to long-term interest rates. Long-duration assets are sensitive to interest rates. KLA's long-term debt weighted maturity is 18.1 years, interest rate 4.67%, locked in. Refinancing risk is close to zero; the rest is all about the numerator story. It's just that with duration rates currently high, $KLAC is having a tough time.