'Get me out': Traders dump software stocks as AI fears erupt Sometimes, amidst major technological waves, the demands on understanding rise dramatically. The primary factors influencing the development of things differ across different periods and historical contexts, making me increasingly aware that nothing is truly immutable. If something is, it might be due to insufficient timeframe or experience, such as within a 30- or 60-year timeframe. For the past two decades, SaaS, characterized by "light assets, high profit margins, and high growth," was the domain of fervent investors. However, since last year, with the accelerated and in-depth development of AI, while Claude Google and OpenAI consistently deliver groundbreaking product launches, SaaS software is being pushed to its grave. The indiscriminate flow of good and bad isn't frightening; what's frightening is structural differentiation. This indicates that there are aspects you haven't seen or understood, where changes are quietly taking place.

Samsung and SK Hynix's combined market capitalization has caught up with Alibaba, whose market capitalization is dragged down by food delivery and e-commerce, and Tencent, whose investment in AI is insufficient and whose gaming business is threatened. So why hasn't Hype Hip-3 been launched yet? SK Hynix and Samsung... SK's earnings call seems to have said they're not doing ADRs anymore.

Both OpenAI and NVDA have reached a point where strategic decisions are needed. The root cause is that NVDA has taken the majority of the FCF (Full Flow Computing) in the entire chain (until 2027), forcing others to fight for survival. From a developmental perspective, they should invest in OpenAI as soon as possible to secure their place in the AI race. Secondly, the entire top AI lab likely only has about 50 talented people? The entire market has been divided up by three companies (Open, Google, and Anthropic). What's even more alarming is that these three can be seen as talents cultivated by Google over the past decade, more familiar with Google's system and style. Therefore, if NVDA wants to gain an early advantage, it can only bet on Open, which has spun off from Google, and I feel it requires a massive investment of hundreds of billions. A major difference between AI and the internet lies in the transmission cycle. From research papers to lab results, to certain application demonstrations, to the factory floor, to the supply chain, to performance, it basically takes about two years. Greg Brockman of OpenAI once said that they desperately wanted to change computing power, architecture, and chip design early on, but no one listened. They were too weak, and changing the entire design and supply chain was too difficult. This led to a slowdown in knowledge transfer. Therefore, to maintain their competitive advantage, they can only start from the very beginning, like acquiring Groq, directly locking in the inference SRAM. So, the opportunity for $AMD$ and $INTC has arrived because the demand is huge.

Professor Kim, the father of HBM, has truly provided a large number of protégés for SK and now $SNDK. Forget about costs, look to the future.

A single thing is influenced by many factors, and different eras have different dominant factors. This makes it difficult to have a single, unchanging investment strategy. If one does, it might be due to an insufficient time frame (e.g., extending to 30-50 years). If the main factors are not firmly grasped or are influenced by noise, it's easy to be swept away by market fluctuations. The impact of AI on silver, copper, and uranium is fundamental and long-term. Understanding the second derivative of investment will inevitably lead from gold to silver and then to current mining stocks. Upward revisions to EPS are inevitable and delayed, ignoring noise. $SCCO Barrick Mining $CCJ $WPM

Elon Musk claims SpaceX will reach 100GW of solar power capacity in three years. Amazon is acquiring copper mines, so why hasn't Musk targeted silver mines in the Americas yet?

Russell 2000 EPS growth forecast of 67% is a bit exaggerated, but I am indeed more optimistic about the Russell 2000. Any growth stemming from EPS is genuine growth. It's secondary factors like liquidity and interest rate cuts that warrant caution. The structural opportunities brought by the new technology wave are everywhere. Claude's evolutionary pace will continue to accelerate, the marginal cost of software development will approach zero, and the monopoly of large companies will be severely challenged, giving numerous startups opportunities. At the same time, we should see three or four companies fall behind in the MEGA 7 this year, their gross margins being swallowed up by energy and electricity. The infrastructure dividends enjoyed by internet giants for over a decade will be gone forever. Who would have thought that one day the balance sheets of these giants would be adorned with electricity and mineral depreciation? A deeper restructuring is still ongoing. There are no old Deng assets, only old Dengs that are out of step with the development cycle of the times.

When researching power equipment, I focused on two areas: gas turbines and transformers. While studying gas turbines in the context of GEVs (Geometric Electric Vehicles), I discovered their extremely long production cycles, with orders booked for up to 28 years. This results in very little supply elasticity. However, they are using aero-engine production lines to modify gas turbines, leading to a substitution effect emerging in the supply chain. Further research revealed many similarities and overlaps in the manufacturing processes of aero-engines and gas turbines, both requiring high-temperature turbine blades, fuel chambers, and so on. In other words, if electricity premiums are maintained long enough and prices remain high enough, is it possible that many aero-engine production lines will shift to gas turbines?

The march of time is a cyclical challenge that everyone must face on their path to growth, and the long-sealed scroll of history is slowly being unveiled. $RKLB Planet Labs $KTOS $ONDS Kraken Robotics

In 2026, software's domination of the world will give way to atoms reshaping it; the main battleground for marginal growth will shift from the scale expansion of the digital world to a revolution in the physical world. We will no longer just update applications; we will begin to update reality.