TradeLeaks

J.P. Morgan on memory: "The order visibility is extending as we progress through the year, but we believe the market is anticipating AI funding risk and power bottleneck driven memory downcycle as early as in 1H27E. This puts memory in a tricky spot as investors debate between "too-cheap to ignore" vs. "it is cheap for a reason." We believe EPS cuts are unlikely to happen in the next 6M as the memory S-D gap is too wide to narrow over a short period of time and memory earnings will have a buffer protected by LTAs and prepayments." $MU $DRAM $EWY $SNDK $SKHY

Various commentary from Qualcomm, Arm, to Meta executives this week during the AI selloff: $LITE CEO: "You had a comment from the CEO of $IREN and I'd second that. There's a massive disconnect in what we're seeing in our demand books. And what the markets are showing. And yet you're seeing a massive, massive selloff." (Fri, Nov 6) $AMD CEO CNBC: "We're at the very beginning of the adoption cycle, where people can see where AI and superintelligence will take us" "We are seeing demand being higher than supply" (Wed, Nov 4) $ARM CEO regarding a debate whether AI infrastructure spending is about to peak: "I think in some states where it gets harder, other states may pick it up to make it easier, but I do believe strongly that if the US hesitates in any way shape performance in terms of data center buildout... "The rest of the world is ready and willing. There is just huge demand for this capacity. And I think globally the US really wants to be able to compete against China. China is not slowing down at all." (Thu, Nov 5) $APLD CEO: Bank construction lending, project financing and bond markets remain available, although interest rates and credit spreads have risen "The big risk is delivering on time." "The demand picture is extremely robust." (Mon, Nov 2) $QCOM CEO: "It's not demand bound. It's a supply bound." "We're short on wafers, we're short on substrate, we're short on assembly. Everything is short because demand is very, very high." "I think what the market is reacting to is what is the speed of the compute build out." $META on political risk: "We did have a governor that won last night who did make data centers a big part of the platform. Say as soon as we get past the inauguration, let's meet because we don't want to lose this investment in our town. We don't want these jobs to go elsewhere." (Wed, Nov 4) Even politicians campaigning against data centers are negotiating. But Meta's president also emphasized that starting new US DCs will become more difficult. TLDR: -> Extremely bullish commentary from accelerator players, networking, and semi companies -> DC developers report strong pricing/demand, but increasingly emphasize financing, permitting and execution risks (more cautious) -> Politicians that previously campaigned against DCs are showing cooperation There's an important difference between demand disappearing and demand being delayed by the inability to build fast enough. Despite the severity of the selloff, I'm not seeing evidence of widespread AI demand destruction in commentary.

ASML is playing DRAM kingmaker as Samsung and SK Hynix compete for ASML's EUV lithography machines, the key to expanding advanced DRAM production, media report, adding the two South Korean gaints are each investing ₩10 trillion to ₩20 trillion won for EUV machines (US$7.2B-$14.4B) for delivery by 2027-2028, but they may have to wait as delivery times stretch - as much as twice as long as the previous maximum - due to heavy demand. ASML said on its Q3 call that the current 2027-2028 supply plan will not meet demand. Samsung and SK Hynix will begin receiving EUV machines in stages starting next year. The one able to ensure on-time delivery will gain a march on the other. In March, SK Hynix disclosed a ₩11.95 trillion EUV equipment order with a delivery deadline of end-2027. $ASML $SKHY $SSNLF #Samsung #semiconductors biz.chosun.com/en/en-it/2026…

It think it's stupid that "AI Bubble" gets used over and over. In infra names you have: - 1-3x fwd P/E memory companies from Samsung to $ESMT. With HBM players now having hyperscaler multi-year price floor LTAs. - Optical companies like $POET traded -$97M under net cash. Valuing the operating business very negatively, not even at $0. - While AI companies with larger premiums continue to record demand over the next 3-5 years. Such as the new $20B LTAs signed w/ $COHR to $5B LTAs signed w/ $LITE. Can someone in the media honestly see negative EV high-beta optical companies, where 2 years of burnrate would still value the company below $0. or single digit memory companies with contracted demand + price floors years out and say that's a bubble? Many of these names are extremely profitable today... But obviously there's other elements inside the theme, such as: - Certain startups like Instinct that look like a private valuation bubble. - Leveraged Neoclouds that didn't finance properly - and AI lab valuations and commitments, which should be viewed cautiously. But everything gets grouped into one with the boogyman of "leverage" or "debt". Yes, valuations can compress or correct from macro, potential regulatory uncertainty, or potential capex reductions. But for the companies physically powering the AI buildout, it looks more like the Great AI Deleveraging of 2026... Than anything like fundamental demand destruction. As more earnings come out and free cash flow accumulates over time. I'm confident in a recovery, especially in bottlenecks, and for markets to eventually discern durability + growth over frothy valuations tied to AI.

Citi: DRAM S/D Outlook $MU $DRAM $EWY $SKHY

AI server maker Wiwynn’s board on Friday approved cash injections of US$250 million for its US unit and $200 million for Malaysia, media report, adding it said customer order visibility remains strong despite recent stock market volatility. October revenue rose 80.9% year-on-year to NT$165.1 billion (US$5.04B), while Jan-Oct revenue hit NT$1.15 trillion (US$35.09B), up 54.0%. The company warned tight upstream materials could still impact monthly revenue. $AMZN $GOOGL $META $MSFT $ORCL ctee.com.tw/news/20261107700…

I found the technical discussion helpful, but investor messaging to be way too conservative. If you look at how $NVDA CEO describes CPO, Jensen states scale up is increasing optical volumes by 15x. $LITE and $COHR are quantifying that by describing $5-$20B agreements spanning 3-5 years out. Having this question, and not talking about Ayar, O-Net, Poet, or ELS partnerships. Then redirecting the question to near term pluggables seems off. The vision also needs work, “communications and sensing” semi company is very unexciting. It’s technically true, but there’s no vision being communicated on being the next $40B optical giant or what Sivers is going to be in the next 5 years that would attract long term holders. The core valuation Sivers gets is from optical opportunities, I also personally disagree that it’s helpful to loop back to former cellular relationships when talking about future $GFS volumes I disagree when you portray the company as “small” when Sivers 100m CW DFB capacity is coming online, alongside large foundry partners. Which is a material % of the world’s annual laser capacity. Investors want to hear about future economic opportunities and how you fill the gap with more qualifications during a bottleneck. Sivers has a tremendous opportunity for CPO/NPO/Pluggables but I’m afraid it’s not being communicated properly. It’s understandable here since EETimes is more technical oriented, but I hope this messaging gets communicated better in the next earnings call.

Samsung, SK hynix See Limited Hit From U.S. Data Center Curbs One person familiar with the industry said chips for data centers already under construction are being produced and that long-term contracts with customers keep demand strong. Policy changes in the United States are unlikely to translate into a drop in chip orders over the short term, the person said. Another industry official said that even if the Democrats tighten rules on AI data centers, demand will continue to exceed supply capacity for some time. "Thirty people are already lined up at a restaurant that is full, and five of them leaving does not create a table," the official said. Even if some data center projects are suspended, supply will still fall short of clearing the backlog of chip demand. $NVDA $AMD $MU $SNDK $AMZN $MSFT $GOOGL

Expert call covering CPO & Optical Modules — CPO deployment pace, switch ASICs vs AI accelerators, 7–8 month optical qualification bottlenecks, pluggable coexistence, and AI training cluster demand. $COHR $NVDA $GOOGL $TSLA $LITE $AVGO