TSMTSMCNeutralTSMC

Yesterday on

TSMC’s October 2026 revenue: NT$586.9B (+14.7% MoM, +59.7% YoY) https://t.co/eOmmay4B7o

NeutralTSMTSMCTSMCAMDASICAtegrity SpecialtyAtegrity SpecialtyNVDANVIDIANVIDIA

Yesterday on

A Taiwanese media report says NVIDIA may cut its orders with TSMC. If that happens, AMD and ASIC chipmakers could pick up some of the wafer capacity currently allocated to NVIDIA. I’ve heard similar rumors. I’ll let you know as soon as I can confirm them. https://t.co/46wCWDA65N

TSMTSMCNeutralTSMCMUMicron TechnologyBearishMicron Technology

3d ago on

1. The counterargument that deliveries can't be pushed back by a year or two just because the two sides worked together from design through validation becomes meaningless once the demand decline is real. 2. Even if they were involved from the design stage, I don't think we have reached the kind of extreme customization seen in foundry, where Micron couldn't do what Samsung was doing. 3. Joint design also constrains which HBM gets used, but I don't think it settles who carries the inventory burden when accelerator demand misses. Say the production plan is cut from 1 million units to 800,000. Is the customer obligated to take the 200,000 units' worth, or can the memory makers simply push that volume onto them? That needs to be clear. The claim that memory has completely shed its commodity character is, in my view, the most dangerous one. There are still more players than in foundry, where TSMC dominates, and they can substitute for one another.

NeutralTSMTSMCTSMCDRAMRoundhill Memory ETFRoundhill Memory ETF

4d ago on

RE: The claim that TSMC will enter memory and disrupt the big three memory makers This doesn't make sense. As far as I know, TSMC has been working on this project for more than five years, but it would be a mistake to assume what they're building is commodity DRAM. TSMC has no DRAM IP, no DRAM patents, and none of the relevant technology. I think the term "3D DRAM" is what's causing the misunderstanding. After talking with industry experts, I believe this is the correct way to read the reports of TSMC entering DRAM. TSMC's earlier published research gives a good sense of the direction. At the 2025 VLSI Symposium, TSMC presented a memory technology based on oxide semiconductors, using a cell made of one transistor and one capacitor (1T1C). The goal was to integrate this memory into advanced logic chips and give processors a cache with low power and low latency. The key feature of the design is that the memory array is built directly into the back end of line (BEOL) interconnect layers of the logic chip. Conventional DRAM is manufactured as a separate chip. To build the memory array in the interconnect layers of an advanced logic chip, TSMC used select transistors with an oxide semiconductor channel and capacitors fabricated with a low temperature process. According to TSMC, the test chip, which integrates memory and logic on a single die, operated at 0.75V. Data retention exceeded 128ms, random read/write cycle time was under 8ns, and endurance was above 10¹⁴ cycles. TSMC positioned the technology as a processor cache that offers higher density than SRAM while keeping power consumption and latency low. So TSMC would be producing 3D DRAM as a next generation technology to replace SRAM. And not in large volumes, either. Anyone saying TSMC will make DDR or LPDDR has no idea what they're talking about.

TSMTSMCNeutralTSMC

6d ago on

Today I met with a veteran investor who argued that memory stocks are not cheap even at a P/E of 3. Below is an edited version of our conversation. Memory LTAs are basically meaningless. The best the LTAs being signed today can do is get customers to shoulder part of the memory makers' capex. Once prices fall, customers will gladly forfeit their deposits (prepayments) and walk away from the LTAs. Historically, whenever prices fell more than 60% from the peak, customers broke their LTAs even after signing them. So even if a 2027 shortage is certain and the stocks trade at a P/E of 3 on 2027 operating profit, 2028 operating profit is uncertain, which makes it hard to justify a higher valuation. He went on to argue that even if 2028 operating profit were certain, we would still have to account for 2029 operating profit. I pushed back and asked why the market doesn't discount TSMC the same way. Why does TSMC get such a high valuation? Wouldn't TSMC also take a big hit if the AI boom fades? His answer: the problem was how violently memory prices rose in the first place. If memory makers had raised prices in a healthy way like TSMC, it might be a different story. But because they hiked prices like a commodity, the market has no choice but to treat memory stocks like commodity stocks. Even if memory makers stay in shortage through 2030, he said, the market will price in the risk of oversupply in 2031. He closed his argument this way: even at 3x 2027 earnings, if memory prices fall in 2028 by enough to break LTAs, that 3x multiple on 2027 works out to 7.8 to 10x on 2028. So current valuations are not cheap.