June 5, 2026
The semiconductor guidance gut check
Beating the quarter is easy. Raising the long view is the new toll.
Hey there, bargain hunter – this week’s AI hardware trade didn’t exactly break. It just got stricter.
For most of 2026, the market has rewarded a simple formula: show strong AI growth, then say the growth lasts longer than people think. This week, the second part did all the work. A solid quarter without a bigger long-term AI target suddenly looked like a problem.
Scoreboard
- Broadcom (AVGO): reported fiscal Q2 2026 revenue of $22.2B (+48% year over year) and non-GAAP EPS of $2.44. AI semiconductor revenue was $10.8B (+143% year over year). It guided fiscal Q3 revenue to about $29.4B (+84% year over year). Shares still dropped hard after results as the company kept its FY2027 AI chip sales target at $100B instead of raising it.
- Marvell (MRVL): fell alongside the group as investors locked in gains in AI networking and custom silicon names. Market commentary around the move pointed to sector weakness and profit-taking after a strong run, with Broadcom’s outlook adding pressure.
That’s the headline: big numbers, big growth, and then a very blunt market response anyway. It wasn’t subtle.
“I said ‘SELL’ before this stock dropped 90%. Today, I’m shouting ‘BUY NOW’ before it’s too late.”
In 2000, Eric Fry told Barron’s magazine that investors should sell a very popular dot-com stock just before it plunged 90%. Today, Eric is saying the exact opposite about it – “BUY NOW!” This same company is now the lifeblood of AI data centers – yet it’s completely undervalued. He says anyone who owns Nvidia stock would be well-served to sell those shares and buy this under-the-radar play instead.
The real reason
Expectations moved faster than guidance.
Broadcom didn’t say AI is slowing down. In fact, the quarter showed acceleration. But investors have been treating the long-term AI target like a scoreboard that should keep getting rewritten upward every few months. When management held the FY2027 $100B AI chip sales goal steady, the market interpreted that as: “maybe the easy upgrades are over.” That’s not necessarily true. It’s just how crowded trades behave when the bar gets too high.
Slight tangent, but it matters: this is what happens when the stock becomes a forecast, not a business. The quarter stops being the product. The future sentence becomes the product.
Why MRVL got caught in it
Marvell sits in the same neighborhood: specialized networking silicon, custom programs, and data center buildouts that are tightly tied to AI infrastructure spending. When a bellwether like AVGO disappoints on the long-term framing, managers often reduce exposure across the whole cluster. Not because MRVL said anything new that day, but because the risk of “good results, bad reaction” suddenly feels contagious.
A major change to U.S. bank accounts could arrive in months.
It may allow closer tracking of your transactions.
My Cheap Investor take
If you own either name, this week is a reminder to separate two questions:
- Is AI demand still real and scaling? AVGO’s AI revenue growth says yes.
- Is the market willing to pay up without a higher long-term target every time? This week says no.
So what now? I’m not treating this as “AI is over.” I’m treating it as a warning that forward-looking targets are now the main trigger for volatility. If you’re adding, think smaller sizing, slower entries, and a preference for companies that can turn AI growth into durable free cash flow, not just exciting percentages.
Worth a look: go back through your watchlist and mark which companies have specific multi-year AI revenue goals versus vague optimism. The market just told us which category it prefers, at least for now.
