Three things happened this week that all point at the same stock, and most traders are only watching one of them.

On Thursday, the Nasdaq Composite fell 1.5% to 25,881.95. Semiconductor stocks were hit hard as well, with the widely followed Philadelphia SE Semiconductor Index down about 3.8% on the session. After that, Bloomberg reported that Alphabet’s Gemini 3.5 Pro is months behind schedule, and Alphabet fell about 4.4% — erasing roughly $200 billion in market cap in a single session. And this morning — options expiration Friday — Chinese startup Moonshot AI announced Kimi K3, a 2.8-trillion-parameter open-weight model it says is competitive with top U.S. systems on coding tasks and other benchmarks. The tape is messy. The signal inside the mess is clearer than it looks.

The stock sitting directly at the intersection of all three events is Broadcom (AVGO).

Where the Numbers Actually Stand

Start with the fundamentals, because they haven’t changed. In fiscal Q2 2026 — the quarter ended May 3 — Broadcom reported total revenue of $22.187 billion, up 48% year-over-year. AI semiconductor revenue came in at $10.8 billion, up 143% YoY, above even the company’s own forecast. Free cash flow hit $10.262 billion in a single quarter. GAAP net income was $9.31 billion. The company’s dividend policy implies a fiscal 2026 annual dividend target of $2.60 per share, representing its 15th consecutive annual dividend increase since initiating dividends in fiscal 2011.

Then CEO Hock Tan guided fiscal Q3 AI semiconductor revenue to $16.0 billion — “over 200%” year-over-year growth — on total fiscal Q3 revenue of approximately $29.4 billion (84% growth). Non-GAAP operating margin was guided stable at 67%.

And the stock fell 15% the day after those results.

That’s the paradox traders need to sit with before putting on any position here. Broadcom posted the kind of numbers most companies won’t generate in their lifetimes, and the market punished it anyway. Why? Because expectations had run so far ahead of reality that even a 143% surge in AI revenue wasn’t enough.

The Setup Right Now — Stocks, Levels, and Why This Week Matters

AVGO hit a 52-week high of $494.18. From there, the stock has pulled back roughly 20-24% depending on where you’re measuring from. As of Thursday’s close, shares were trading near the $370-$390 range — down again on the semiconductor sector rotation, this time compounded by the Alphabet Gemini news.

Slight tangent, but it matters here: Alphabet is widely considered one of Broadcom’s most important custom silicon customers, given Broadcom’s long-running involvement with Google’s TPU program. When Alphabet falls 4.4% on a Gemini delay and investors start questioning whether Google is losing ground in the AI race, that’s not just a Google story. It raises questions about the capex trajectory of one of AVGO’s anchor clients.

Then add Kimi K3 this morning. Moonshot AI released what it’s calling a 2.8 trillion-parameter open-weight flagship with a ~1 million-token context window, with API pricing listed at $15 per million output tokens (and $3 per million input tokens, plus a lower cached-input rate). That’s enough to force a repricing of “who can ship frontier-ish performance, and at what cost,” even before the market decides how reproducible the benchmarks really are.

Here’s what that means for Broadcom specifically: if open-weight models start commoditizing frontier AI capabilities at dramatically lower price points, the hyperscalers running massive custom silicon programs may eventually face margin pressure on their AI products. That’s a third-order risk at best — but the market tends to price tail risks quickly and ask questions later.

The Bull Case for Custom Silicon Doesn’t Go Away

What’s interesting is that none of the above changes the core structural story for AVGO. The custom ASIC thesis is actually strengthened by the Kimi K3 announcement, not weakened.

Here’s the logic: if inference costs are collapsing and open-weight models are flooding the market, hyperscalers aren’t going to cut their AI infrastructure spending — they’re going to accelerate it. The economics of running more inference at lower cost requires more compute, not less. That’s precisely why Amazon, Google, Meta, and Microsoft are spending aggressively on custom silicon programs. General-purpose GPUs are increasingly giving way to purpose-built ASICs that run specific model architectures more efficiently at scale.

Broadcom’s Jalapeño chip — co-developed with OpenAI and taken from initial design to tape-out in just nine months — is the clearest evidence of this trend.

The tension in the story isn’t the technology. It’s the valuation overhang.

Valuation, Sentiment, and What the Street Is Saying

AVGO carries 44 buy ratings among covering analysts and zero sell ratings, with a consensus price target of approximately $523. The 52-week low is $260.80. At current prices near $380-390, the stock trades at a roughly 40% discount to its 52-week high. The trailing P/E sits near 63x — still elevated, but compressing rapidly as the stock has pulled back and earnings continue to grow.

The forward setup hinges entirely on one thing: whether Broadcom delivers on the $16.0 billion fiscal Q3 AI semiconductor guidance. If Tan comes to the fiscal Q3 earnings call — expected around early September — and reiterates or raises the “line of sight” toward more than $100 billion in AI semiconductor revenue in fiscal 2027, the setup for a re-rating becomes straightforward. If that number gets cut, or if a major hyperscaler signals capex pullback, the multiple compression story has further to run.

Wall Street’s confidence in the name is high. The sell-off since the June earnings is being characterized broadly as macro-driven multiple compression, not a fundamental deterioration story. Fed Chair Kevin Warsh’s “higher for longer” posture is squeezing premium multiples across the board, and Broadcom — which trades at a significant premium to the broader semiconductor industry due to its AI positioning and VMware software component — has above-average sensitivity to real rate moves.

Sector Context: What’s Actually Moving the Tape

Thursday’s session told a specific story about where institutional risk appetite is heading into Q2 earnings season. The S&P 500 fell 0.5% even as more stocks rose than fell within the index. The Nasdaq Composite fell 1.5% to 25,881.95.

Consumer Staples (XLP) rose 2.9%. Healthcare (XLV) outperformed. REITs held up. This is a classic rotation signature — institutional money de-risking out of high-multiple AI infrastructure names and moving toward defensive sectors with lower valuation risk.

The 10-year Treasury yield sits at 4.56%. The bond market is still pricing in rate hike odds for September — close to 60% probability for at least a 25bp move, per CME FedWatch. That’s a headwind for every name in the semiconductor complex that trades above 40x earnings. Broadcom is no exception.

PPI data released this week showed headline producer prices at 5.5% annually, down from 6.0% in May, with a 0.3% monthly decline. That’s directionally favorable, but not enough to change the Fed’s calculus meaningfully. The July rate hike odds dropped sharply after the CPI print earlier this week — from 42% to 17% — but the broader trajectory remains hawkish under Warsh’s leadership.

Scenario Modeling

Bull Case

Broadcom delivers fiscal Q3 AI semiconductor revenue at or above $16.0 billion, with Hock Tan raising or extending the fiscal 2027 AI revenue ambition. Hyperscaler capex commentary from Amazon, Google, Meta, and Microsoft in their upcoming Q2 earnings calls reinforces AI infrastructure spending durability. The Fed signals a pause rather than a hike at the September meeting. In this scenario, AVGO re-rates toward the $450-$500 range as the Q3 guidance print restores institutional confidence and multiple compression reverses. The custom ASIC thesis — bolstered by the Jalapeño partnership with OpenAI — becomes the dominant narrative again.

Base Case

Broadcom hits fiscal Q3 guidance broadly in line, reiterates the 2027 ambition without raising it, and the stock consolidates in the $370-$420 range through the summer. Macro headwinds from the rate environment keep premium multiples under pressure, but the fundamental AI infrastructure story holds. The stock grinds higher slowly as earnings growth catches up to the valuation. Analyst consensus targets drift modestly lower but remain well above current price levels.

Bear Case

A major hyperscaler — potentially Google, given the Gemini delay — signals reduced custom chip orders or extends contract timelines. The Kimi K3 / Chinese AI competition narrative accelerates, causing broader questions about Western hyperscaler AI ROI. The Fed hikes in September, real yields move higher, and high-multiple tech names see a second leg of multiple compression. AVGO breaks below the $340-$350 support zone and retests the $300-$320 range, potentially approaching the 200-day moving average near $317. This would represent a 35%+ drawdown from the peak.

Active Trader Strategy Framework

The setup here is not a single-catalyst trade. This is a multi-variable decision framework requiring specific conditions before sizing into exposure.

• Key resistance levels: $420-$430 zone — this is where the 50-day EMA and recent overhead supply cluster. A sustained close above this level on volume would signal the post-earnings flush has stabilized.
• Key support levels: $370-$380 near-term; $340-$350 as the critical structural support. Below $340, the technical picture deteriorates materially.
• Volatility context: The VIX has been climbing this week, and options expiration today may amplify intraday swings. Position sizing should account for above-average realized volatility in AVGO specifically.
• Catalyst calendar: Hyperscaler Q2 earnings begin rolling in over the next two weeks — Amazon reports July 30, Meta July 29. Any commentary on AI capex spending trajectories will move AVGO more than most macro data points. Broadcom’s own fiscal Q3 report isn’t until early September, which means there’s a multi-week window where the stock trades on sentiment and sector flows rather than hard numbers.
• Risk management: The combination of a hawkish Fed, Chinese AI competition, Gemini delay overhang, and a high trailing multiple creates a risk profile that rewards patience. Scaling into positions rather than full commitment at current levels is the more defensible approach given the number of unresolved variables.

The part people skip in the Broadcom debate: the locked-in multi-year hyperscaler contracts and the “over 200%” fiscal Q3 AI semiconductor guidance aren’t abstract projections. They’re the near-term signposts the market will trade against. Whether the stock re-rates toward those fundamentals in Q3 or Q4 depends almost entirely on whether the macro environment allows premium multiples to expand again.

That’s the question that doesn’t have a clean answer yet. The numbers argue for patience. The chart is still resolving. And the next two weeks of hyperscaler earnings may be the most important data points for the custom silicon trade that traders have seen all year.