The AI Bull Market Starts Marking to Market
US Stocks · Options · News · Views
Yesterday the market was still debating AI stock-picking.
Today the market delivered a reminder: not every AI story keeps going up.
All four major indices fell, and the S&P's nine-day win streak snapped. Energy and healthcare held up; consumer, financials, and tech lagged noticeably.
On the surface this looks like a normal pullback.
But what matters isn't how much the index dropped.
It's that the AI chain is moving from "telling stories" into "verifying earnings."
Broadcom's print wasn't a blowup — but the market wasn't satisfied.
That's the most important thing to watch today.
This AI rally isn't over. But it's no longer in the buy-with-eyes-closed phase.
The market was willing to pay for AI imagination. Now it's asking a more brutal question: with all this money going in, can it actually come back out?
Whoever can prove AI spend converts to revenue keeps feasting.
Whoever is just riding emotion, valuation, and hot money needs to worry about sudden profit-taking.
Today's US selloff wasn't just a technical breather.
The Middle East headlines kept flipping, oil got bid up (helping energy outperform), then the headlines eased and oil's gains trimmed — a he-said-she-said tape.
The data didn't offer much comfort either.
ADP payrolls rose from 109k to 122k, above the 117k expectation.
May ISM services PMI rose from 53.6 to 54.5, also above consensus.
The problem: the prices index stayed elevated, rising from 70.1 to 71.3.
Translation: services demand is intact, but cost pressure hasn't gone away.
Fuel, energy, tariffs, transport, raw materials — these are slowly dragging inflation back into the picture.
So the market's biggest fear right now isn't an immediate economic deterioration — it's that inflation won't come down and the Fed won't be able to pivot cleanly.
On the numbers already reported, Broadcom's print wasn't bad.
Q2 revenue grew 48% to $22.2B, slightly above consensus. AI semiconductor revenue came in at $10.8B, also ahead of expectations. Adjusted EPS of $2.44 beat too.
But what the market really stared at was the forward guide.
Q3 AI semiconductor revenue guide of $16B came in below the $17.2B Street consensus.
Total revenue guidance, while above sell-side consensus, missed the higher bar set by some buy-side estimates.
That's why this isn't an "earnings bomb" but the market still reacted violently.
The AI trade's biggest fear isn't that it isn't making money today.
It's that the future growth rate stops getting revised up.
For the past several quarters, the market had grown accustomed to Broadcom constantly nudging AI expectations higher. This time, it did not raise the forward bar.
That's tantamount to telling the market: don't rush to discount the entire future.
Broadcom's issue isn't really a Broadcom-only issue.
What it pulls into view is the central question of the entire AI chain: can Big Tech's AI spend keep going?
Broadcom has long-term partnerships with Google, Meta, and other large customers.
But a multi-year agreement isn't the same as revenue recognized each quarter.
An agreement is an agreement. Cash landing is cash landing.
If the customer's spending pace slows, downstream suppliers' revenue realization slows with it.
That's the biggest risk in the AI chain right now.
If Big Tech can earn enough from AI, they keep spending, and the whole chain keeps feasting.
But if AI monetization can't keep pace with capex, the only options are to slow spend, issue debt, or even raise equity.
At that point, downstream isn't feasting anymore.
It's leftovers, hot water, and maybe even losing the bowl.
The contrast between AVGO and MRVL captures the current market state.
In 2025, when the market chased custom AI silicon, Broadcom was the consensus leader. Capital looked at customers, orders, earnings, and fundamentals; AVGO ran hard while MRVL went sideways for a year.
But over the past two months, the logic flipped.
MRVL is running faster — not necessarily because its fundamentals suddenly dominate AVGO's.
The bigger reason: it's smaller, and easier for flows to push.
Short-term money isn't looking for the most reasonably valued name.
It's looking for what's easiest to lift, what has the least overhead supply, what can manufacture the most sentiment.
That's a classic flow-driven tape.
You can participate — but don't mistake it for a fundamental re-rating.
Short-term money has dates, not marriages.
When it's hot, the passion is real. When the tide goes out, it doesn't look back.
CRWD's numbers weren't ugly.
Adjusted EPS of $1.10 on revenue of $1.39B both beat.
Full-year revenue and EPS guidance were also slightly above consensus.
The company also approved a 4-for-1 stock split.
A split lowers the per-share price, improves tradability, and makes options look cheaper — drawing in smaller tickets.
But a split is not a fundamental catalyst. It's arithmetic.
One share becomes four; market cap is unchanged; shareholder equity is unchanged; the company's earnings power is unchanged.
CRWD's real issue is valuation.
It had run too far too fast, so post-print profit-taking is no surprise.
Pulling back to first support doesn't end the trend.
But it does illustrate: high-valuation names that fail to keep beating expectations become easy targets for profit-takers.
Over the coming week, the market has another major variable: the SpaceX listing.
This won't necessarily top the market immediately.
But it could reallocate capital.
If SpaceX lists at a sky-high valuation, the market may not look at valuation seriously at first.
What hot money trades is float scarcity, scarce supply, and concentrated sentiment.
In other words, they're trading not whether the company is worth it, but whether enough people are scrambling for shares in the short window.
That raises a question: where does the money come from?
Likely from the high-elasticity AI names that have ripped recently and are sitting on thick profit tapes.
MRVL, CRWD, and a whole cohort of high-valuation growth names could all become sources of cash-out flow.
So the faster a stock has been chased lately, the more cautious you should be around the SpaceX listing window.
NFLX has fallen for eight straight sessions, conspicuously weak. The better observation zone isn't 79–84 but below 77. The real value opportunity is whether a strong stabilization structure emerges somewhere inside the wide 60–77 range.
ISRG has entered the watch zone below 405, but reaching 405 doesn't mean buy immediately. 321–405 is a very wide tolerance band; only a strong stabilization structure justifies action.
TSLA's key level is holding the uptrend line. Reclaiming 428, ideally back above 433, eases the short-term tape. A break of 404–405 effectively invalidates the prior false breakout structure.
MSFT, AMZN, NVDA have all pulled back into observable ranges, but that doesn't mean load up in one shot.
MSFT: a break below the current zone may put it back into a chop structure; smaller positions can scale in, heavier positions don't need to force adds.
AMZN: the aggressive zone is close, but the safer buy is closer to 224 or below, ideally near 219.
NVDA: from 203 down is observable, but the market cap is so large now that what it needs most is volume contraction on the decline. Even on further weakness, as long as volume slowly fades, that's not a bad thing.
IGV's break below 101 interrupted the software squeeze. It's not bearish — it's back inside the 90–107 chop zone, and needs to rebuild energy.
The most important thing right now isn't to panic at every down print, and it isn't to chase anything labeled AI.
The market is still inside the AI mainline, but the rhythm has changed.
The prior phase was capital willing to pay for imagination.
The next phase will be capital demanding companies prove delivery.
So strategically: core positions can keep watching the long-duration thesis, but high-elasticity short-term trades need active profit defense.
The faster a stock runs, the more you must separate fundamental drive from flow drive.
If it's the former, give it time.
If it's the latter, don't confuse short-term sentiment with long-term faith.
Over the next one to two weeks, watch three things: whether AI leaders stabilize on declining volume, whether SpaceX siphons off hot money, and whether IGV and high-valuation growth names keep bleeding.
This isn't an untradable market.
But remember one line: the AI bull market is still alive — the margin of safety has gotten thinner.
Long-press the QR code to follow the official account for daily US market coverage and views.
Disclaimer: This article is personal market observation and trading review only, and does not constitute investment advice. Markets carry risk; trade with caution.