Steven Philips on Premarket: Apple is green - 'We know who the real heavyweight is. On the “Mag7” seesaw when six go down, one goes up! 🙂'
on Premarket: Apple is green - 'Nice idea but not feasible for the same reasons that Apple doesn’t pursue making its own memory: From Gemini: “Converting an existing Intel logic fab to produce DRAM memory chips would be a massive, time-consuming undertaking that could take nearly as long as building a brand-new fab—likely two to three years or more. While Intel does have its own memory intellectual property, it would almost certainly need to license a chip design or form a joint venture to manufacture standard high-volume DRAM or HBM. The timeline and licensing requirements break down as follows: ⏳ Conversion Speed & Timeline Converting a logic fab (designed for CPUs and GPUs) to a DRAM fab is incredibly complex due to radically different manufacturing steps. • The “Cleanroom Overhaul” Barrier: Logic chips favor high-speed logic transistors with complex, multi-layered metal routing. DRAM requires massive arrays of high-aspect-ratio capacitors stacked vertically or deeply etched into the silicon to hold electrical charges. • Equipment Swap: Intel would have to rip out or heavily modify its current toolsets. On tech forums like Reddit, users have highlighted that the efficiency loss and sheer capital required mean converting a logic fab takes almost as long as constructing a new one from scratch. • The Ohio/SK hynix Scenario: Recent developments show that rather than converting an active facility, Intel has been in talks with South Korea’s SK hynix to either lease out space or form a joint venture at Intel’s upcoming Ohio facility to kickstart U.S.-based DRAM and HBM lines. This leverages a “blank canvas” rather than a disruptive conversion. Licensing Requirements Intel would face steep patent and design hurdles if it attempted to go it alone: • The IP Deficit: Intel famously exited the DRAM market in the 1980s to focus on microprocessors. Major memory players like SK hynix, Samsung, and Micron hold a tight web of foundational patents, especially around High Bandwidth Memory (HBM). • The Need to License: To produce industry-standard DRAM quickly, Intel would need to license memory IP technology. This is a core reason behind their ongoing discussions with SK hynix. • Intel’s Proprietary Workaround: Intel has filed patents for its own alternative technology, called XBM (Cross-Batch Memory). XBM is designed to bypass traditional DRAM fabrication by using an open-source interconnect standard (UCIe) to stack memory directly on current logic nodes. However, to build traditional planar DRAM or standard HBM, licensing remains a necessity.” “Converting a modern U.S. logic fab to a DRAM memory fab is financially impractical. It would cost an estimated $4 billion to $6 billion, which represents roughly 40% to 50% of the cost of building a brand-new facility from scratch. Because semiconductor equipment is highly specialized, a conversion is not a simple reconfiguration. It requires a massive write-down of existing logic assets and a complete capital expenditure (CapEx) overhaul. Cost Breakdown of a Fab Overhaul The total cost is driven by the structural mismatch between logic and memory manufacturing: • Tool Replacement ($3.0B – $4.0B): • Logic fabs prioritize complex multi-layer metal routing (requiring specific lithography and etching tools). • DRAM requires specialized, high-aspect-ratio deposition and etch equipment to build deep vertical capacitors. • Intel would have to scrap or relocate billions of dollars in logic tools and purchase entirely new memory-specific fleets. • Cleanroom & Facility Reconfiguration ($500M – $1.0B): • Altering the sub-fab utility piping, ultra-pure water systems, and chemical delivery lines to match the new tool layout. • Licensing & Technology Transfer ($500M+): • Upfront IP licensing fees or joint-venture buy-ins (e.g., with a partner like SK hynix) to secure the blueprints and proprietary manufacturing processes (recipes) for modern DRAM/HBM nodes. The “Hidden Cost”: Billions in Lost Revenue The biggest financial deterrent isn’t the equipment—it’s the opportunity cost. • Zero Revenue for 24+ Months: Intel would have to take a multi-billion-dollar logic fab completely offline. • Asset Write-Offs: Intel would be forced to take massive, immediate depreciation and impairment charges on the logic equipment being ripped out, severely damaging its quarterly earnings and balance sheet. • The Co-Location Solution: This extreme financial penalty is exactly why Intel’s rumored strategy is to provide cleanroom space to memory makers at new sites (like Ohio) rather than converting old ones. Co-location allows Intel to generate foundry revenue immediately without destroying its existing logic capacity.” Just not practical, plus memory prices would likely have fallen by 2030-2031 when the new fab comes online, and DRAM pricing wouldn’t make enough back to justify it.'
on Apple's John Ternus talks touchscreen Mac (with product video) - 'Auto incorrect: “*buying* me a new laptop”'
on Apple touches 351.43, closes at $348.91, a record high - 'In AH trading Apple has been up as much as $3.50. Wild stuff.'
on Apple touches 351.43, closes at $348.91, a record high - '@David I don’t think so. The earnings report is on Monday so there is likely a pause on Apple till then. Also the election is Tuesday which will likely overshadow any other news. Also, the AI stocks have been hit making Apple more of a safe haven. Google was down over 12% today. Intel was down ~20% after hours following a poor earnings report. I believe I heard that Meta did not report very well (could be wrong). Looks like there is enough angst out there to keep Apple afloat for a while. Plus the daily barrage of good news about new products coming out can’t hurt. Just one man’s opinion.'
on Apple's John Ternus talks touchscreen Mac (with product video) - '”The premium is nice but is the upgrade cycle shorter for iPads than for MacBook Pros?” I remember when the iPhone remained on the market for 15 months while Apple moved its launch date from July to September. Who’s to say Apple isn’t doing something similar next year?'
on Apple's John Ternus talks touchscreen Mac (with product video) - 'Nah, probably is true…if you list it on eBay or similar.'
on Apple touches 351.43, closes at $348.91, a record high - 'So what’s the pattern? Tomorrow we’ll get some FUD, possibly Duo or Macbook Pro supply chain cuts, then some negative ‘color’ by Gurman, to take us into Monday’s earnings call. After that call, there’ll be a drop when all the “buy rumor/sell news” people cash out, and likely a slow decline with increasing holiday supply FUD. Then we’ll start to hear positive news based on actual holiiday sales, with a January run-up. There’ll be cashing out by the speculators who bought during this run-down, but the actual number of shares will continue to decline thanks to buy-backs and the buy-and-hold crowd. What am I missing?'
on Apple touches 351.43, closes at $348.91, a record high - 'In AH aapl touched at least $351.98 even though yahoo shows $361.67 as the AH high. How do I know? Because I woke to a pleasant inbox surprise of a confirmation of 351.98. Cheers to the longs indeed.'
on Premarket: Apple is green - 'After GOOG yesterday, it’s AMZN today, the earnings of which seem to be spooking investors. You could see in AH activity, AMZN going down and simultaneously AAPL going up. Somebody’s waking up to the reality of mega Capex spending.'
on Premarket: Apple is green - 'Here’s an Apple investment I’d like to see – while Intel is not a state-of-the-art foundry, neither is the manufacture of memory chips. I’d like to see Apple do a multi-year deal with Intel to supply a certain percentage of their memory chip needs at a below current market price. Fabbed in the USA, of course. Even a below market price would still be highly profitable for Intel, and would buy them time to reestablish their mojo. It might also just give Apple a bit more leverage in negotiating contracts with their current suppliers if they perceive that a “hungry” Intel is coming for their bacon…'
on Apple's John Ternus talks touchscreen Mac (with product video) - 'Gregg, The premium is nice but is the upgrade cycle shorter for iPads than for MacBook Pros? Besides that, this should be a nice boost for Apple’s Mac revenues once this has a full quarter to take effect.'
