Live Recap 3: Inside the Memory Supercycle — NAND, DRAM, and the AI Compute Shift
1.Live Review Introduction
Tiger Brokers livestream hosted by Vyann, featuring Ross Dong, Founding Partner at Morning Cloud Asset Management and founder of Gongxing Academy. With over 15 years of experience as an equity trader at firms including J.P. Morgan and KCG, Ross holds a Bachelor of Science in Applied Mathematics from Columbia University. His research spans macro trading, U.S. equities, and the structural dynamics behind the global AI memory supply chain, and he currently manages over $30M in AUM while leading a community of 15,000+ members. Memory stocks including $SK hynix(SKHY)$ have pulled back sharply since late June even as the AI infrastructure buildout keeps accelerating, leaving investors to weigh whether this is a short-term top or a buying opportunity within a longer boom.
Disclaimer: All content shared is purely for investor education purposes and does not constitute any financial advice, investment recommendation or trading signal. All trading involves significant risks. Please conduct independent research before making investment decisions.
2. A Different Kind of Upcycle
This memory cycle differs fundamentally from previous ones because the demand driving it is structural, not cyclical, powered by AI rather than the usual PC or smartphone refresh cycles. DDR5 adoption is accelerating rapidly across hyperscale data centers, and memory content per server continues to climb — reinforcing that this isn't a typical inventory-driven swing.
3. Pricing Momentum
Both NAND and DRAM pricing continue on an upward trajectory. Data center memory pricing rose roughly 25% in the third quarter — a step down from an even sharper second-quarter increase, but still a clear continuation of the broader uptrend rather than a reversal.
4. The Undersupply Timeline
Annual capacity growth for high-bandwidth memory (HBM) is running around 20-30%, while AI-driven demand growth is closer to 200-300% — a gap wide enough that the undersupply isn't expected to meaningfully ease until 2028 or 2029. Memory makers have historically been conservative about capex given oversupply risk, but stronger demand visibility for HBM specifically is giving them more confidence to invest now.
5. Kimi K3 and the AI Angle
Moonshot AI's newly released Kimi K3 is the largest open-source AI model globally at 2.8 trillion parameters, with a 1 million token context window — drawing comparisons to the market impact of DeepSeek R1's release. Open-source models tend to be more memory-intensive than frontier closed models, since longer context windows and growing key-value cache requirements increase dependence on HBM and storage even as quantization techniques improve, reinforcing memory demand rather than undercutting it.
6. Optical Bottlenecks Reinforce the Thesis
Optical valuation multiples hit record levels in June and early July, reflecting a shift in the AI infrastructure bottleneck: as GPU clusters scale up, optical interconnects increasingly become the limiting factor. Despite aggressive capacity expansion from optical component makers, supply chain checks still point to a structural shortage in premium components — a dynamic that supports pricing power industry-wide.
Closing Takeaway
From the memory chips themselves to the AI models consuming them to the optical infrastructure connecting it all, the underlying supply-demand imbalance looks set to persist well beyond this year. Finally, what this all means specifically for $SK hynix(SKHY)$, plus highlights from the live audience Q&A.
7. Risk Reminder
High-volatility AI and memory-related stocks carry substantial trading risks. Market participants without sufficient foundational knowledge are advised to complete education modules before initiating live positions.
8. Post-Event Resources
Viewers can access further insights via Ross Dong's official website, tmipartner.com, or his YouTube channel, TMI Partner. You can also follow him on X as Ross Dong, or via Tothemoon as Ross美股投资. The full livestream replay is available on the Tiger Trade app. Community participants may share demo trading plans and market observations tagging TB Live to join bonus activity rewards.
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
