Semiconductors are up — will AI tokens follow? The answer might surprise you.

The U.S. semiconductor index surged nearly 4% in a day, with Astera Labs soaring 16%.

AI-concept stocks saw a V-shaped reversal, and the bulls are celebrating.

But on the other hand, GigaDevice has openly said that the prices of its memory products will see a “considerable drop.”

What’s rising is chip design; what’s falling is memory pricing.

Within the same big AI cycle, the upstream is partying, while the midstream is cutting prices.

So what about AI tokens?

Coins like FET, RNDR, and TAO—do they move with U.S. semiconductor stocks, or do they go their own way?

My answer is: follow in the short term, don’t follow in the medium term, and judge by fate in the long run.

First, the short term—yes, and they’re following closely.

On June 29, the U.S. semiconductor index jumped nearly 4%, and AI-concept stocks exploded across the board. During the same period, the total market cap of AI tokens posted a weekly increase of 11%, outperforming the overall crypto market’s 4%. RNDR jumped 14% directly after news from NVIDIA’s GTC.

On the surface, AI tokens and U.S. semiconductor stocks look like “brothers in arms.”

But the truth isn’t that simple.

This round of AI token gains is more about “narrative premium” than “fundamentals moving together.”

What does that mean?

U.S. semiconductor stocks are up because companies like NVIDIA and Broadcom truly have profits, orders, and production capacity. AI tokens are up because the market believes, “AI is so hot—there should also be an AI-related asset in crypto.”

One is backed by real performance; the other is riding a trending story.

A story can be told for a while, but it can’t last forever.

Now look at the medium term—divergence has already begun.

Over the past week, Wall Street hedge funds sold U.S. information technology stocks at a pace that set a historical record. A Morgan Stanley analyst even directly warned that the semiconductor sector could be topping out temporarily.

The Philadelphia Semiconductor Index has risen 155% over the past year and 106% from the start of the year. An index that has doubled—how many more times can it double?

Meanwhile, clear signals of loosening prices on the memory side are already unmistakable. GigaDevice said a “considerable drop”—this isn’t a minor adjustment; it’s a trend shift. Apple and Microsoft previously raised prices together due to rising chip costs, and end consumers are already paying for the expansion of computing power. Now that memory prices are falling, it means pressure from upstream costs is starting to ease—but it also means the most high-profit phase for AI hardware may be coming to an end.

So how about AI tokens?

South Korea’s $5,180,000,000,000 AI chip plan is pulling risk capital out of the digital asset space at an unprecedented speed. This isn’t pocket change—it’s national-level capital diversion.

On one side, institutional funds are withdrawing from semiconductors; on the other, state-level investment is draining crypto.

AI tokens are stuck in the middle, pleasing neither side.

So my judgment is:

Short term (1-2 weeks): AI tokens will follow the sentiment swings of U.S. semiconductor stocks. When semiconductors rise, AI tokens will tag along to “drink the broth”; when semiconductors fall, AI tokens will be the first to get smashed.

Medium term (1-3 months): AI tokens will most likely underperform U.S. semiconductor stocks. Because U.S. semiconductor stocks have earnings support, while AI tokens only have narrative. When the market starts asking, “Where is your GPU? Have you rented out your computing power?” the story can’t be told anymore.

Long term: it depends on two things—

Whether decentralized computing power can truly produce a closed-loop business model, rather than staying stuck at the “testnet TVL” stage.

Macroeconomic liquidity. If the Federal Reserve continues to sound hawkish, all risk assets will be killed off—AI tokens will die the fastest.

Finally, here are three lines to send you:

“U.S. semiconductor stocks are up because they’re really selling chips.

AI tokens are up because they’re selling dreams.

Dreams have a much shorter shelf life than chips.”

“When memory begins to drop,

it doesn’t mean the AI cycle is over—

it means the phase where you can make money with your eyes closed has ended.”

“Is capital rotating from AI stocks into crypto assets? Don’t make a dream.

The reality is the opposite—crypto assets are being drained by AI.”

My strategy (pure sharing, not advice):

I will participate in AI tokens in stages, but I will never go all-in or hold them long-term.

I’ll take a quick bite of the sentiment premium and run—no “falling in love” with them.

Because I know—

When semiconductors start to diverge, memory starts to drop, and institutions begin to retreat,

the best days for AI tokens may already be behind us.#TradFiCFD黄金大师赛 #Saylor暗示增持BTC $BTC $TAO $WLD