A rough observation: the dollar’s peg is quietly extending from oil to AI computing power. Before, it was “buying oil requires dollars”; in the future, it might be “accessing top-tier AI requires dollars.” Holding dollars, in a sense, means holding access to the world’s leading AI.

To break this monopoly, one path is accelerating the offensive with domestic large models—but more fundamentally, it would be the emergence of a non-dollar-dominated AI ecosystem and settlement network. As for stablecoins, the mainstream ones right now are actually dollar on-chain stand-ins. Don’t count on them for now.

What does a truly “non-dollar AI ecosystem” require?

At least three conditions are simultaneously met:

Model layer: It has world-class large models, with performance that doesn’t lag behind the US top players.

Compute layer: It has self-controlled AI chips and a compute network, and does not depend on Nvidia.

Settlement layer: It has a cross-border payment network, allowing you to buy compute and AI services using non-USD currencies (such as RMB, euros, or a basket of currencies).

Right now, only China is pushing all three of these things at the same time, but none of them has reached a “globally available” level yet.

Let’s talk about the outlook for each sector as well.

The crypto sector is still going through waves of elimination. BTC/ETH remain more mainstream. With high volatility, leverage and “air projects” will continue to be cleansed. In the future, mainstream coins will gradually take over the market, and the upside will also gradually shrink. Digital gold’s volatility will likely stabilize, and at the end of the bear market I will continue to hold.

The “seven sisters” in US stocks have shifted from broad-based gains to differentiation. The story is still about AI. Only AI businesses that can deliver cash flow and realize profits can keep making new highs. Microsoft, Google, and Nvidia—continue to hold. Apple—hold at least until this year’s earnings/publicity event.

The Nasdaq/S&P 500 long-term core trend is upward, but the path is more bumpy. The S&P 500 has better defensiveness, while the Nasdaq is more driven by interest rates and the tech cycle—I will continue to DCA and hold. The AI/HBM-driven memory cycle points upward. Among SanDisk, Micron, and Hynix, these three are the most solid choices within Micron-related plays. As for cyclical stocks, don’t tell a “perpetual growth” story when sentiment is at a high point; set up proper stop-losses and trade through volatility—you can’t hold forever. Personally, I believe there will be a phase peak; I’ll hold Micron in the medium term. The most favored sector right now is NewCloud. Cloud-based AI compute is a deterministic trend. Cloud providers’ capital expenditures still have resilience, and leading platforms should benefit more. On the NBIS side, the backlog of orders is enough to carry them for a long time; short-term holding.

Regarding gold and silver: the logic behind central bank gold buying, the rate-cutting cycle, and de-dollarization hasn’t broken. Gold is in a long bull run, but the wealth-creation effect is slow—the most狂热 period has already passed. Silver has higher leverage, but the position weight relative to profit is too low, so I don’t hold. For crude oil, concerns about premium versus recession are pulling back and forth; most likely it will trade in a range-bound pattern, with the center of gravity slowly drifting downward. I would short it on rallies and don’t hold.