Unusual Intelligence

Kimi K3 has received far more love than we expected, and our GPUs are feeling it. Over the past 48 hours, demand has pushed close to the limits of our current capacity. To protect the experience of existing subscribers, we're temporarily pausing new subscriptions and prioritizing compute for current members. Existing subscribed users are not affected. We're adding capacity as fast as we can and will reopen new subscription spots in batches. Going forward, we'll also split membership into two more focused plans: Kimi Membership for Kimi Web, App, and Work; and Kimi Code Membership for coding workflows. This will help us match compute more precisely and keep the experience stable. Thank you for your patience and understanding!

Corporate Insiders are dumping stocks at the 2nd fastest pace in more than 20 years 🚨 🚨

4 big things happened this week and if the AI/Memory trade breaks down from here, we may point to: -$AAPL asked for access to Chinese Memory -$META pivoted to selling “excess” compute -$PLTR CEO ripped Open AI/Anthropic as thieves -Open AI may sell stake to USG

I occasionally follow your posts. They come across as knowledgeable and as though you have invested a great deal of time in your research, which is certainly commendable. However, it sometimes seems to me that, by making the narrative increasingly elaborate, you are trying to convince not only your followers, but also yourself. Regardless of how rational the market may or may not be, it is still showing us that money has been flowing out of the sector for some time. The moves are significant, and each daily close sends a very clear signal to anyone who genuinely knows how to read price action. Until that changes, all theories will continue to fall apart, because talking about how attractively valued something is does not mean much until the price itself shows that it has reached a level the market is willing to accept as a sustainable equilibrium. Naturally, with every further decline it appears cheaper, but that is precisely where the greatest psychological trap lies, the trap that 90% of people fall into.

A good comprehensive article covering major topics in memory supply demand debate. Having seen five cycles already, I have seen industry analysts failing to predict top and bottom every time. Equity analysts couldn't see the biggest bull run just a year ago when Micron was trading below $80. How can we say that they are not wrong about their predictions of the near term now. Supplg side is predictable. But there are many unknown unknowns that are impossible to anticipate on the demand side of AI. These unknown unknown can be related to technology disruption or it might be simple economics of hyperscalar's capability to spend. At some point, the focus would be on the RoI of these investments. Anything can happen. My advise. Don't put all your eggs in memory. Play it safe.

I came across an interesting analysis today. It argued that while the unwinding of leveraged positions has contributed to the recent decline in memory stocks, the market is also beginning to price in the prospect of future supply expansion. The argument goes as follows. Even if the memory shortage persists throughout 2027, research firms, sell-side analysts, and other prominent industry observers broadly agree that the supply-demand imbalance will begin to ease in 2028. Given that Samsung Electronics and SK hynix have announced massive fab expansion plans—and that other memory manufacturers are also racing to expand capacity—it seems reasonable to conclude that the market has already begun pricing in the increase in supply expected from 2028 onward. There is a well-known rule of thumb when it comes to memory stocks: share prices tend to anticipate the peak in memory prices by about two quarters. But something I have long wondered—and perhaps deliberately tried to ignore—is whether that anticipation necessarily has to be limited to two quarters. Is it really impossible for the market to price in the peak three or even four quarters in advance? I found this analysis extremely interesting. More importantly, I found it quite convincing.