https://x.com/brad_setser/status/1976691234103058664?s=46

Aligned w our theory

Changed my mind, exited $UAMY @ $12.2-$12.35 and re-added some positions that were derisked earlier back.

Reasoning is if the Chinese were trying to find maximum leverage for trade negotiations, they’d first try to push DT to the point where he has no more tolerance.

This would explain why they were trickling in subtle yet escalatory news over the past week. Now DT losing it on Truth Social today is probably near that point.

The Chinese know that if they continue pushing, DT will not back down and they risk potential backfire. I believe both parties have very strong incentives to not let the trade war blow out of control again, like per April.

Hence, if the Chinese have found their maximum pressure point, the next logical step is to soften their hold in exchange for trade benefits (maybe xx% tariffs reliefs instead of full reliefs)

All just my speculation. Not sure, am also trying to avoid over pontificating about macro and the thought processes of megalomaniacs.

China's clarification on rare earth controls; no retaliation to USA 100% tariffs, more of a "we're not banning rare earth, but just reviewing controls" tone set. Imo, at best a softening of stance, at least, not escalatory. Seems like an outreach to DT for him to deescalate graciously. Will need to see how DT perceives it though.

"Going forward, the Chinese government will conduct reviews in accordance with laws and regulations, grant licenses to eligible applications, as well as actively considering the applicability of facilitation measures such as general licenses and license exemptions to effectively promote legitimate trade. I want to emphasize that China’s export controls are not export bans. All applications of compliant export for civil use can get approval, so that relevant businesses have no need to worry. The Chinese government will work with all countries as always, to firmly safeguard world peace and stability in neighboring regions, and jointly maintain the stability of global industrial and supply chains."

"Willful threats of high tariffs are not the right way to get along with China. China’s position on the trade war is consistent: we do not want it, but we are not afraid of it. China urges the U.S. to promptly correct its wrong practices, adhere to the important consensuses of the phone calls between the two heads of state, protect the hard-won outcomes of consultations, continue to use the China-U.S. economic and trade consultation mechanism, and address respective concerns and properly manage differences through dialogues and on the basis of mutual respect and equal-footed consultation, so as to ensure the stable, sound and sustainable development of the China-U.S. economic and trade relationship. If the U.S. insists on going the wrong way, China will surely take resolute measures to protect its legitimate rights and interests."

https://x.com/MOFCOM_China/status/1977193939729104997

Just in — so far, so good. Fingers crossed though, still some time till market opens.

https://x.com/zerohedge/status/1977385782030557621?s=46

Thank you for your attention to the matter!!

Added $763.HK (ZTE) and $981.HK (SMIC) longs at HK$43.23 and HK$80.21 respectively. I’ll name this theme the China Decoupling Beneficiaries.

While I believe tensions should ease heading into the Trump–Xi meeting in November, the Pandora’s box has already been opened. Both superpowers have now revealed their cards and the leverage they hold over each other. This dynamic will likely accelerate efforts on both sides to decouple critical chokepoints in their respective supply chains.

On the surface, they may start to play nice in a bid to buy time — as neither can yet fully do without the other — but beneath that veneer lies a sleepless and tedious grind toward independence from each other. At the eye of this storm is simply AI. AI is a national security imperative, and neither side can afford to fall behind in the race, nor risk being held hostage by export controls from its rival.

Cut $ALAB @ $209 for -6% on OpenAI <> Broadcom’s AI accelerator deal.

The AI accelerators will use AVGO’s ethernet-based switches, likely scale up ethernet (SUE). With AVGO’s stronghold on custom silicon, their networking architecture will likely become the industry’s primary architecture.

This is not good for $ALAB’s PCIe-based switches as I think it’s a winner-take-all market. ALAB’s valuation is also on the high side and much of that is dependent on the growth of their switches.

With the amount of excess overflow coming out of OpenAI — as seen from $AMD and $AVGO lately — I think $INTC may potentially catch an OpenAI deal too.. maybe something to do with their idle advanced packaging capacity.

Also exited $MU long @ ~$190 and added to $SNDK long @ ~$126

Strong set of results from $TSM

- double beats on rev and eps
- price/wafer crossed record high again
- all end-markets up DD% q/q
- “AI demand is insane”. Mgmt’s conviction on AI demand is incremental
- planning on building out more capacity to meet future demand; including acquiring second U.S. site in U.S.
- talking to customers’ customers (vs just customers previously) to understand AI market requirements

Overall demand picture remains really healthy all around, across all end-markets (and likely to be the case through 2026). Capacity is also expected to remain relatively tight. If there was ever a good time for someone else to start another external foundry..

Moonshot play $TGEN finally starting to move after a long consolidation.. +35% from avg cost of $9.06 to $12.25

Moving on no news though is kind of strange. Not sure if it’s insider front running or just a small cap rerating. Hopefully $VRT just buys it out one day.

Cut $MRVL @ $89.5 for +2% gain

Shifted $MRVL allocation to re-add $MU long @ $199.56.

St is finally acknowledging the idea of a lengthened memory cycle. UBS just upgraded the stock, let’s see if that’s enough fuel for the stock to power through ATH

BOOM

Rumoured to be $MSFT Maia 2 chips. If this is accurate, it’s IFS’s first external customer. History happening!

$MU and $SNDK crossed ATH again last night, with $MU closing @ $208.75 and $SNDK @ $149.64, +23.8% and +30.5% respectively from when we first went long.

Meta also announced yesterday at OCP that they were prioritizing QLC NAND adoption in its data centers to reduce costs, which may have been a driver. All things considered, the supply picture for DRAM/NAND appears to be very tight through 2026, pointing to an extended memory cycle. I guess the big qn on investors' mind however is whether "this time its different?". Could the excess driven by AI spend change the pre-AI cycles of memory?

Tbh, I have no clue given my generalist background. I try to approach highly cyclical companies with more caution. However, it's also worth noting that we have barely moved into inference demand, and hence the current memory/storage demand could just be the beginning of many future AI drivers that will be exponentially more memory intensive.

From Seagate's earnings rephrased: "in the entire last 150 years, 15 billion images were created. With AI, that same number of images were created only in the past 1.5 years. The resolution of the images are incredibly high as well. Beyond that, a typical 1 minute video consumers 100x more data than one high-def image."

Excluding image and video generation (which is an incredible driver on its own), I believe to make AI truly useful (imagine GPT as a PA), it needs to have context on who you are and what you want. No one wants a PA that acts like a new-hire everyday. That requires memory. As the ability of agentic AI scales, more tasks can be handled autonomously and data can be processed faster — that requires memory/storage. For robotics to be truly useful, we need to collect and process petabytes of data from the real world with cameras and sensors everyday — that's memory/storage. For edge AI to be fast, data needs to be duplicated and stored at locations closest to where its consumed — that requires storage.

All in all, I think the long-term case for memory is very strong. Along the way however, there will be pullbacks as I think the rally will be similar to climbing a wall of worry. Any small fluctuations in memory spot pricing, no matter how short term it is, will dent confidence. However, I think the pullbacks will be BTD oppys.

Re-added a $RDDT long @ $205.36. Stock has retraced 20% from when we exited at $257.

Think r/r is decent now to give the upcoming print a shot — given that the DAU growth concerns are now known and “priced in”. Also partly a rotation move away from AI infrastructure as mentioned yesterday.