Has Memory Bottomed?

Only 2.2 percent of US households currently hold paid AI subscriptions, a level the speakers compare to e-commerce at roughly 2 percent of US retail sales around 2004 to 2005, which has since grown to approximately 17 percent. The speakers predict paid AI subscriptions will eventually reach 50 percent of US households and 50 percent of global households, with revenue flowing to Micron, Microsoft, Meta, Google, and other hyperscalers. They argue AI adoption will move faster than internet adoption because the hardware barrier is absent: users need only a phone or computer they already own, making the main bottleneck compute supply rather than device distribution.

Memory trade stocks are down 30 to 40 percent from their highs. The speaker had previously recommended re-entering at that level while acknowledging a further 10 to 15 percent downside was possible, and stocks did fall an additional 10 percent after that call. The speaker advised reducing positions when Micron rose 11 percent post-earnings and then retraced the move, recommending investors raise cash and rotate away from memory stocks to wait for a reset, and specifying this approach should apply only to a smaller allocation rather than an entire portfolio.

A margin call event in the Korean stock market illustrates the conviction and risk embedded in the memory trade. Approximately 26 billion dollars in margin loans were outstanding in Korean equities, and when SK Hynix and Samsung declined roughly 25 percent, 1.2 million accounts hit margin calls, 300,000 accounts were liquidated, and forced selling totaled approximately 1.5 to 2 billion dollars, denominated as 2.3 trillion won. The speakers interpret the scale of that borrowing as a sign investors felt strong or delusional confidence the memory and AI trend would continue uninterrupted. Jonah describes Micron and SanDisk as volatile, not yet priced to perfection, but offering a low Sharpe ratio with high return on capital if held through drawdowns. He also frames the US-China AI arms race as structurally bullish for component makers regardless of which nation wins, noting Chinese model makers have largely caught up to frontier AI and the Chinese government has announced significant domestic AI funding.

Gold is described as off 26 percent from its highs but having defended the 4,000 level for approximately six weeks, which is characterized as technically strong. The conservative price target is cited as around 4,800, the previous resistance level, and gold miners are described as a good bet. The recommended trade is to buy gold with a tight stop at 4,000, selling immediately if that level breaks because the thesis would be disproven, while a move above 4,300, the yearly moving average, is cited as a safer entry signal for more cautious investors. Jonah states that the number one driver of gold is central bank accumulation and that is literally all that matters, while separately acknowledging he does not fully understand what drives gold despite having headed a precious metals desk. The broader argument is that technical analysis is more appropriate for gold than equities because gold has no fundamentals, only flows, and its reflexive nature makes the 4,000 support level critical.

The speakers flag nuclear and uranium as a less crowded trade relative to AI and semiconductors, with the US having signed a deal with Saudi Arabia to help build nuclear plants, extending the uranium investment thesis. A company called Centris is mentioned as involved in running reactors and enriching uranium and has pulled back significantly in price, though the speaker has no position yet and is still researching whether Centris would be involved in foreign construction under the US-Saudi agreement.

On Iran, the speakers expect conflict to continue in cycles of provocation and retaliation, with regime change described as the only path to a genuine end. A reported Israeli intelligence operation involved cultivating Ahmadinejad after he was pushed out by the Ayatollah, but the plan collapsed when Ahmadinejad got cold feet, failed to reach a safe zone, was captured, and is now out of the picture. The speakers assess that the US and Israel likely miscalculated, believing they could install a friendlier Iranian leader more quickly. The speakers predict the Strait of Hormuz will lose strategic relevance within ten years as East-West pipelines are built on the west side of the Gulf, permanently removing Iran's primary geopolitical leverage. Oil long-end prices are not rallying hard because markets view the Strait of Hormuz threat as short-term, and the speakers predict Trump will manage escalation cycles by releasing tension whenever markets are threatened, preventing a major sustained spike in oil prices.

On crypto, the speakers suggest it may have temporarily bottomed and that if memory goes nowhere for one to two weeks, rotation back into gold and Bitcoin is likely. The odds of the Clarity Act passing are described as ticking back up, with passage considered bullish for crypto.