NextFin News - Investors are heading into chip earnings week in a far less forgiving mood than they were a month ago. The PHLX Semiconductor Index has fallen more than 11% from its June peak, and on July 28 it briefly dropped nearly 6% intraday as traders cut exposure to the same names that powered the artificial-intelligence rally. The immediate question is no longer whether chip demand exists. It is whether the market has decided that the easy part of the AI trade is over.
The Selloff Is About Expectations, Not Demand
The move matters because it is happening after a long run-up, not after a disaster. Semiconductor shares had become one of the market’s most crowded momentum themes, and by late July the group was being repriced as investors asked a harder question: if revenue still rises, what happens when the market stops paying for every dollar of growth as if it were guaranteed to compound forever? That is the key tension running through earnings week.
The selloff also became a broader market event. On July 28, market coverage described a nearly 6% drop in the PHLX Semiconductor Index that briefly pulled the Nasdaq-100 into correction territory. That is a sign that the trade is no longer confined to one stock or one earnings miss. It has become a cross-asset judgment on the durability of AI capital spending, the sustainability of valuations, and how quickly investor sentiment can shift when the bar for proof rises.
The calendar is loaded with catalysts. Intel said it would report second-quarter results on July 23. AMD said it would report fiscal second-quarter results on Aug. 4. Broadcom already posted fiscal second-quarter revenue of $22.2 billion, up 48% year over year, with AI semiconductor revenue of $10.8 billion, up 143% year over year. ASML raised its 2026 sales outlook to €43 billion-€45 billion and guided to gross margin of 54% to 56% after reporting second-quarter sales of €9.3 billion and net income of €2.9 billion. NVIDIA’s investor-relations calendar points to its next quarterly results later in the season. In other words, the market is not waiting for one print. It is waiting to see whether several prints still validate the same narrative.
That is why the sector’s reaction cannot be dismissed as a routine breather. If earnings were still being read through the lens of pure AI scarcity, a 48% revenue gain at Broadcom or a raised full-year outlook at ASML would likely stabilize the group. Instead, those numbers have become the standard against which every future quarter will be judged. The first-order effect of the selloff is lower prices. The second-order effect is a harsher regime for valuation and capital allocation across the AI supply chain.
The shift is visible in the way investors talk about the trade. A few months ago the question was whether AI demand was real. Now the question is whether it is profitable enough, fast enough, and broad enough to justify the capital intensity of the supply chain that serves it. That is a very different debate. It turns chipmakers from a story about scarcity and growth into a story about payback periods, backlog conversion, and how long customers are willing to keep spending before they demand results.
That context matters because the global semiconductor industry still sits in what many forecasters describe as an expansionary phase. Gartner said worldwide semiconductor revenue is projected to exceed $1.3 trillion in 2026, a 64% increase, with memory revenue expected to triple. That is a remarkably strong baseline. It also means the current selloff is not about a collapsing industry. It is about whether the market has already capitalized too much of that growth into share prices.
Why This Feels Cyclical in the Short Run, but Structural in the Long Run
My call is that the current move is cyclical in the near term and potentially structural underneath it. Cyclically, chip stocks often overshoot after a strong run. They sit at the intersection of growth, inventory, and sentiment, which makes them highly reflexive. History offers several reminders: the 2018 semiconductor downturn was driven by inventory and trade fears, the 2022 collapse reflected a broader growth reset and tightening financial conditions, and the 2024 AI rotation showed how quickly enthusiasm can outrun near-term shipment data. In each case, the group reversed violently before finding a new base. That pattern still matters here because the sector had become crowded and because the July selloff looks like de-risking after extended gains.
But the structural question is harder to dismiss this time. The AI trade now depends on a chain of decisions by hyperscalers, cloud providers, and enterprise customers, and that chain is being judged on return on invested capital rather than on narrative alone. Broadcom’s AI semiconductor revenue of $10.8 billion and ASML’s raised outlook show that demand is real. They also prove the market now has hard numbers, not just hopes, to work with. That changes the pricing mechanism. A stock can rally on any sign of accelerating adoption when expectations are low. Once expectations get high, the same adoption must also prove margin durability and cash conversion.
That is the second-order effect the market is focused on. The first-order effect of AI capex is more demand for semiconductors and tools. The second-order effect is pressure on multiples if investors conclude that spending is being pulled forward faster than it is being monetized. The third-order effect is portfolio rotation: money leaves the most expensive AI names and moves toward sectors that offer either lower valuation risk or less dependence on the same capex cycle. That is why a chip selloff can matter beyond chips. It can reshape leadership inside technology itself.
The setup also exposes the sector’s concentration. When a few large names dominate index performance, a small change in sentiment can force leveraged and passive investors to reduce exposure across the whole group. That makes the tape more fragile than the fundamentals alone would suggest. It is one reason a sector that still reports strong sales can trade as if it is in trouble: the market is not only discounting profits, it is discounting positioning. The July move is therefore not only about what companies will say in earnings calls. It is also about how much exposure investors are willing to keep if those calls merely confirm the already-impressive consensus instead of extending it.
The strongest bullish counter-thesis is straightforward: nothing fundamental has broken. ASML still sees €43 billion-€45 billion of 2026 sales. Broadcom’s AI business is still growing at triple-digit rates. Intel’s second quarter beat expectations on revenue and margin. If those trends continue, then the selloff is likely just a mid-cycle reset caused by crowded ownership and profit-taking, not the start of a lasting bear market for chips. That argument is not weak. It is exactly the right counter if one believes the AI spending cycle is still in its steepest phase and if one thinks the market has simply run ahead of the next few quarters.
What would falsify the structural-bearish version of the story? A broad reacceleration in guidance across the major chip designers, equipment makers, and hyperscaler customers, plus evidence that free cash flow and margins are expanding alongside revenue. More concretely, if the next two earnings cycles show rising order visibility, better margin guidance, and no slowdown in capital spending from the biggest AI buyers, then the July pullback would look like a cyclical washout rather than the start of a lower-multiple regime. The burden of proof is now on the bulls, but the rebuttal is measurable.
“We now expect total net sales for 2026 to be between €43 billion and €45 billion, with a gross margin between 54% and 56%,” ASML said in its second-quarter results.
“In our fiscal Q2 2026, total revenue reached a record $22.2 billion, up 48% year-on-year,” Broadcom said on its earnings call.
Those figures are precisely why the selloff is more interesting than a generic risk-off move. If demand were fading outright, those numbers would not still be arriving. Instead, the market is reassessing how much of the next leg has already been priced in. That is why chip stocks can slide even while the industry’s revenue outlook remains robust: the market is moving from scarcity pricing to proof pricing.
There is also a broader market signal embedded in the move. When a sector becomes one of the primary contributors to index leadership, the repricing of that sector can pull risk appetite lower even if the rest of the economy is not deteriorating. That is what gives the selloff its leverage. It is not just about a handful of semiconductor companies; it is about whether investors are still willing to pay for long-duration growth when the discount rate on future AI profits is rising in practice, even if it is not rising on the policy line.
That helps explain why the current debate is less about whether AI demand exists and more about how much of the next three years has already been pulled into today’s price. A cycle can remain strong and still become a bad trade if the market paid too much too early. That distinction is central here. The business may keep growing. The shares may still underperform. Those are not the same question.
One more layer matters: the market’s response to chip earnings now feeds back into the rest of technology. If semiconductor leaders report solid numbers but their shares still fall, software and cloud names can lose the valuation support that comes from assuming infinite downstream demand. If they report strong numbers and shares rebound, that same feedback loop can restore confidence across the broader AI complex. In that sense, earnings week is not only a test of chipmakers. It is a test of whether the technology sector still has a single dominant narrative or whether the narrative is beginning to fragment into winners, beneficiaries, and expensive exceptions.
That fragmentation is what makes the current period more dangerous than a simple dip in a favorite theme. In a straight cyclical correction, sentiment clears, valuations reset, and the same stocks lead again after a short break. In a structural reset, the market starts to demand a higher quality of proof, a lower tolerance for overbuild, and a clearer path from spending to shareholder returns. The difference between those two regimes can be small in price terms at first, but large in what the market rewards afterward.
What Earnings Week Can Still Change
The next catalyst is not simply whether results come in above or below estimates. It is whether management teams can show that AI-related revenue is still converting into durable profit pools. If that link holds, the sector can recover quickly because the recent decline has already reduced some of the excess optimism. If the link weakens, the market will likely conclude that chipmakers deserve a lower multiple even while revenue is still growing.
That split matters by time horizon. In the short term, volatility can remain elevated because positioning is crowded, sentiment is fragile, and any guidance that fails to clear a high bar can trigger fresh selling. Over the medium term, the key question is whether leading-edge chips, networking gear, and manufacturing tools keep seeing order growth fast enough to offset any cooling in the broader AI spending frenzy. Over the long term, investors must decide whether AI infrastructure becomes a normal industrial capex cycle or a genuinely new profit regime. Those are different outcomes. A normal cycle can still be profitable, but it usually earns a lower multiple than a structural monopoly on growth.
The beneficiaries in the near term are the parts of the market that can offer earnings growth without forcing investors to underwrite the most expensive AI assumptions. The exposed names are the chipmakers and equipment suppliers whose valuations already assume several years of uninterrupted expansion. If earnings week confirms broad demand and strong monetization, the July slide will look cyclical and mean-reverting. If it shows that investors were right to question return on capital, the move will look more structural, and the burden of proof for the next rally will be much higher.
The cleanest signal to watch is whether the next round of guidance emphasizes accelerating orders and expanding margins more than the need to keep up with spending. If the balance shifts toward monetization, the sector can stabilize. If it shifts toward capex without payback, the market will keep cutting exposure. The market’s verdict will not be about AI’s existence. It will be about whether AI’s profits arrive fast enough to justify the bill.
Investors are not bailing on chipmakers because they doubt AI exists. They are bailing because they no longer want to pay for AI as if every dollar of spending will come back twice.
Explore more exclusive insights at nextfin.ai.
