On July 21, after the close of the U.S. stock market, the chip sector, previously stirred up by Kimi K3, finally caught its breath.

The market's previous concern about the "DeepSeek 2.0 moment" did not further escalate. Although SK Hynix still closed down 1.86%, previous star stocks like SanDisk and Micron have turned upward, showing signs of stabilization and rebound. The entire semiconductor sector, after experiencing more than a week of intense selling, seemed to have found a temporary balance point.

Why was it that this time Kimi K3 did not replicate the panic script of DeepSeek 1.0?

1. K3's Narrative Has Changed: It's Not "Sufficient Computing Power," But "Computing Power Shortage"

DeepSeek 1.0, which created a huge stir last March and April, had a core narrative of "low cost, high efficiency, and no computing power shortage"—it trained a nearly top-level model using very few GPU resources, directly challenging the investment logic that "AI performance depends on frantically piling chips."

But Kimi K3's story is entirely different.

Although K3 shocked the industry with 2.8 trillion parameters and extremely low single inference costs, another consequence following its release is equally noteworthy: computing power shortage. The popularity of K3 exceeded expectations, causing the inference infrastructure of the dark side of the moon to come under rapid pressure, making capacity expansion an urgent matter.

What does this mean? It means K3 is not proving that "not so many chips are needed," but is proving that "even if the model efficiency is high, the rate of demand growth will always outpace supply." This narrative of "computing power shortage" is precisely the most favorable support for the chip sector—it reassures investors that the demand for AI chips will not disappear due to improved model efficiency, but may actually continue to expand due to the explosive growth of application scenarios.

The logic of DeepSeek 1.0 is "efficiency substitutes scale," while the logic of K3 is "efficiency releases demand." These two narratives have diametrically opposite effects on chip stocks. This is also the fundamental reason why "DeepSeek 2.0" did not unfold.

2. Google's Game-Changer: Embedding Gemini into Chips

At a critical moment of seeking direction in the chip sector, Google has thrown out a wild card that could change the rules of the game.

Alphabet is developing a brand new AI server chip, code-named Frozen v2. The design concept of this chip is extremely radical: directly embedding part of the architecture of the Gemini model into the silicon chip itself.

This is not the traditional concept of "optimizing software for hardware," but rather "baking the model blueprint into the chip"—by reducing the movement of data between computing units and memory, dramatically lowering the power consumption and latency for each inference.

Google's engineers expect Frozen v2 to achieve astonishing energy efficiency: the number of tokens processed per unit of power could be 6 to 10 times that of the current most advanced Ironwood TPU. In comparison, Ironwood is already Google's seventh-generation TPU, and it merely doubled performance per watt from its predecessors. The generational leap of Frozen v2 far exceeds any previous chip upgrade by Google.

After the news broke, Alphabet's stock price briefly rose from around $350 to around $359 during trading. The market is clearly reassessing Google's long-term competitiveness in the AI infrastructure field—if Frozen v2 can indeed be commercially deployed by 2028, Google will possess one of the most efficient large model inference infrastructures in the world, making its cost advantage in AI services difficult for competitors to replicate.

However, the significance of Frozen v2 extends far beyond "Google's own business" for the entire chip sector.

It conveys a key signal: the AI giants have not slowed down their investment in hardware due to increased model efficiency. On the contrary, they are pushing competition toward a deeper, more customized dimension—dedicated chips (ASICs). From Nvidia's general-purpose GPUs to Google's dedicated TPUs, and now to embedding model architectures directly into silicon with Frozen v2, the competition in AI hardware is shifting from "who has more cards" to "whose cards are smarter, more efficient, and more specialized."

This shift means that the demand for AI chips will not shrink because "models have become smaller and cheaper." Quite the opposite, the trend toward specialization and customization will create a more diverse and segmented chip demand—this is a long-term benefit for the entire semiconductor industry chain.

3. Earnings Season Exam: The Next Week Decides the Direction

Although the chip sector has temporarily stabilized, the test is not over.

In the coming week and more, core players in the AI industry chain will successively release their quarterly earnings reports. The significance of this earnings season is extraordinary—the market is not only looking at whether the numbers are good, but also at several key questions that determine whether the AI narrative can continue:

For Google, Meta, and Microsoft, will capital expenditures continue to burn? Can revenues generated by cloud services and AI services cover the increasingly heavy depreciation, leasing, and power costs? If not, the narrative of "AI monetization" may experience cracks, and the turning point for capital expenditure growth may arrive sooner than we think.

For SK Hynix, can the money burned by cloud vendors ultimately translate into the ability to raise prices on storage chips, expand market share, and increase profits? Hynix needs to prove in its earnings report that it is a real money-making link in the AI industry chain, and not just a "middleman making a margin."

What we really need to watch for this round of earnings is not simply "how high capital expenditures are," but rather "whether they can continue to exceed expectations." After several consecutive quarters of "surprises," the market's threshold for exceeding expectations has been raised to an extremely high level. Any marginal slowdown—whether it is guidance downgrades, softened language, or key data falling short of expectations—could spark a new round of selling.

4. In Conclusion: In the Eye of the Earnings Season Storm, Options are an Important Anchor

The chip sector is experiencing a highly uncertain window period. Bulls say that K3's computing power shortage and Google's Frozen v2 prove that the demand for AI hardware is far from peaking; bears argue that the growth rate of capital expenditures is about to peak, valuations have long been overstretched, and earnings reports are difficult to continuously exceed expectations.

In this environment of information fragmentation, the risk of betting on a single direction far outweighs potential gains.

BIT's options functionality will officially launch this week, providing perfectly matching tools for this "uncertain direction and volatile" market environment:

• Hold chip stocks + buy put options: Insure positions before earnings reports, locking in downside risks

• Single-direction buy call/put options: Bet on post-earnings direction using funds significantly less than the underlying stocks, with maximum loss limited to the premium

• Simultaneous buying on both directions: Uncertain whether earnings reports will be a surprise or a shock? Bet on both sides, as long as volatility is great enough, profit is possible

The storm of earnings season is approaching. In an uncertain market, those who have options are the ones who are entitled to speak calmly.

Risk Warning: The market trends, valuation calculations, and product descriptions mentioned in this article are for reference only and do not constitute investment advice. Trading in U.S. stocks and their derivatives involves market volatility, leverage, and liquidity risks; short selling may face the risk of unlimited losses; options trading carries the possibility of total loss of premiums; past performance does not indicate future returns. Investors should make prudent decisions based on their own risk tolerance and consult professional investment advisors if necessary.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。