#Gate股票暢聊季 Nvidia loses $1.4 trillion in a month, but I cast my vote for Broadcom

After the US midterm elections in early November, the market began worrying about tighter data center permitting and regulation. The AI sector saw its most severe single-day sell-off of 2026: SOXX fell 12% in one day, Nvidia dropped more than 9%, and Broadcom fell more than 10%. The first rebound came on November 6 (AMD +11%, Intel +8%, Micron +5%). On November 9, the Philadelphia Semiconductor Index rose another 6.85%, with AMD up 13.26%, Broadcom up 8.78%, and Micron up 6.25%—but Nvidia bucked the trend, closing down 4.27% at $178.67, wiping out about $193 billion in market value. Since its record high on October 6, Nvidia has lost nearly $1.4 trillion in market value in a month.

My view: this is a double hit from valuation and policy, not a collapse in demand

Look at three sets of hard data:
AMD posted record third-quarter revenue of $13.4 billion, with data center business nearly doubling year over year and net income up 95%; its fourth-quarter guidance is still moving higher;
Broadcom posted revenue of $22.2 billion last quarter, up 48% year over year, including AI chip revenue of $10.8 billion, up 143%; in June, it reiterated its target of more than $100 billion in AI revenue for FY2027;
The four major cloud providers are expected to spend a combined roughly $700 billion in capital expenditures in 2026, with some analysts expecting that figure to exceed $1 trillion in 2027.

Demand is fine. What’s broken is the “policy path” and valuation multiples. On November 5, Nvidia’s forward P/E briefly fell to 15x, its lowest level in nearly three years and just half its long-term average of 30x—the market has already priced in the idea that “this is the earnings peak.”

So, on “will the rebound continue or will there be another pullback?” my answer is: most likely, a wide trading range. There won’t be a real trend before two dates—the US CPI report on the evening of November 10, and Nvidia’s earnings on November 17 (the company’s guidance is $108 billion ±2%).

If I could pick only one stock, I’d choose AVGO (Broadcom)

Three reasons:

① It directly benefits from “de-Nvidia-ization.” Broadcom makes custom ASICs developed in-house by hyperscalers, as well as Ethernet switches. The more customers like Google, OpenAI, and Anthropic want to move away from a single GPU supplier, the firmer Broadcom’s orders become. This business is locked in by contracts, not PPT expectations.

② In the inference era, cost-effectiveness favors ASICs. Training is about peak compute; inference is about cost per token. As agents take off at scale and inference volumes rise exponentially, custom chips with lower unit costs will continue to gain share. This is a structural trend, not a quarterly fluctuation.

③ After the drop, the risk/reward is better. After a single-day decline of 15%, it has priced in “policy panic” more thoroughly than Nvidia, but its revenue mix doesn’t carry a “single-product category faith premium.”

The downside is clear too: very high customer concentration. If even one of its largest customers shifts its capital spending, Broadcom’s exposure cuts both ways.

How I see the other three

NVDA: A 15x forward P/E paired with 70% growth isn’t bad in terms of risk/reward, but there have recently been non-operational headwinds at the stock level, which will amplify short-term volatility. The real answer will have to wait for management’s comments on FY2028 on November 17—that matters far more than a single quarter’s numbers.
AMD: It has the most upside sensitivity, and its data center business really is nearly doubling, but the 13% single-day rebound has already priced in expectations. Chasing the rally offers worse value than waiting for a pullback.
MU / storage: The cycle is the most certain, with HBM in short supply through 2027, but this segment has already run up substantially this year. I don’t think it will be “the next round’s leader”; it’s more of a “confirming play.”

The next round depends on four signals

1️⃣ Whether CPI cooperates (the interest rate path determines valuation upside)
2️⃣ Nvidia’s FY2028 guidance on 11/17, and whether US domestic customers have started cutting orders
3️⃣ How much the four major cloud providers raise their 2027 capex forecasts in their Q4 earnings reports
4️⃣ Data center permitting: will local officials in Texas, Ohio, and New York soften their stance?

This isn’t the end of the AI story; it’s a reshuffling of who gets pricing power in the inference era. My answer is Broadcom: the best risk/reward, the strongest narrative, and the least dependent on faith.

The above is a review of public information, not investment advice. Markets carry risk; manage your own positions 🙏$AVGO