This AI selloff looks more like a positioning reset than the start of a dot-com-style collapse
The market is having the wrong AI argument. This week’s drawdown in semis and infrastructure looks less like a thesis break and more like a crowded-trade unwind in a group where demand, margins, and order visibility are still showing up in the numbers.
The cleanest way to read this AI selloff is not as the start of a dot-com replay, but as a reset in a trade that had become too crowded and too easy to own. The June 23 drop gave bears a neat historical analogy, and Jeremy Grantham’s bubble warning gave that analogy a headline. But analogies are not fundamentals. What matters is whether the companies at the center of the AI buildout are still converting demand into revenue, earnings, and visible enterprise spending, and the latest prints say they are.
The dot-com comparison breaks down on the most important point: today’s leaders are not being valued on hope alone.
NVIDIA
is trading at 32.97x earnings while growing revenue 65.5% with a 63.0% net margin.
Micron
sits at 25.10x earnings with 48.9% revenue growth and a 55.9% net margin. Those are not the fingerprints of a sector built on vapor. They are the numbers of companies already monetizing a capex cycle that customers are still funding.
Micron’s latest quarter is the strongest rebuttal to the idea that AI demand is suddenly rolling over. The company reported revenue of $41.46 billion, up from $23.86 billion in the prior quarter and $9.30 billion a year earlier, while management explicitly tied results and outlook to the strategic value of memory in the AI era. That matters because memory is not a sentiment trade; it sits inside the real hardware stack. If one of the most direct beneficiaries of AI server demand is printing that kind of acceleration, the burden is on bears to show a demand break, not just a scary chart.
The tape itself also argues for a positioning reset more than a collapse. On June 23, the Nasdaq fell 2.2% while the Dow was down just 0.1%, with selling concentrated in AI-linked names. Micron dropped 13.2% that day. That is exactly how crowded leadership unwinds tend to look: the market does not reject risk broadly, it de-grosses the winners everyone owns. A true thesis break would look wider, messier, and more fundamental than one ugly session that hits the same cluster of semis and infrastructure names at once.
That does not mean every AI stock deserves a pass. The better frame is to separate earnings-backed infrastructure from the more aggressively priced second tier.
NVDA: 32.97x P/E, 65.5% revenue growth, 63.0% net margin
MU: 25.10x P/E, 48.9% revenue growth, 55.9% net margin
AVGO: 44.90x P/E, 23.9% revenue growth, 38.8% net margin
AMD: 113.88x P/E, 34.3% revenue growth, 13.4% net margin
VRT: 64.40x P/E, 27.7% revenue growth, 14.4% net margin
That spread is the whole story. There is no single AI bubble multiple. Broadcom at 44.90x earnings is not cheap, but it is also delivering record revenue, operating profit, and free cash flow while citing accelerating AI semiconductor demand and a 67% non-GAAP operating margin. AMD, by contrast, is carrying a much richer 113.88x earnings multiple with a thinner 13.4% net margin. The market should be more selective here, and this selloff may actually be doing that work.
Bears are right about one thing: parts of the trade are stretched, and when a theme gets this consensus-owned, valuation stops being a footnote. Arista at 49.72x earnings and Vertiv at 64.40x are not priced for disappointment. If enterprise AI spending slips or cloud capex pauses, those names can re-rate fast. But that is still a different argument from saying the whole AI complex is a dot-com bubble. A valuation reset in the expensive edges is not the same as a collapse in the core demand engine.
And the core demand engine still looks tangible. Broadcom’s recent messaging that the AI experimentation phase is over and that private cloud deployment is becoming the real enterprise venue matters because it points to actual workload migration, not just pilot-project excitement. Its June 24 product announcement with OpenAI and its platform tied to more than 20 gigawatts of global AI deployments through 2028 add to that sense of order visibility. We do not need to pretend every management team is infallible to see the difference between speculative storytelling and multi-year infrastructure commitments.
This is why the market is having the wrong argument. "Bubble or no bubble" is too blunt to be useful. The sharper question is which companies are funding AI enthusiasm with real earnings power and which ones are merely renting the narrative. On that test, the leaders still look more durable than the selloff implies. Even NVIDIA, with a modest 1.9% YTD gain despite its enormous $4.66 trillion market cap, does not look like a stock levitating solely on momentum. It looks like a giant digesting expectations while fundamentals keep doing the heavy lifting.
Our verdict is that this week’s air pocket looks more like a healthy purge of crowded positioning than the first leg of a dot-com-style collapse. The evidence is not that AI is cheap; it is that the most important parts of the stack are still posting the kind of growth, margins, and management commentary that point to live demand rather than broken demand.
What would change our mind is straightforward: a real deterioration in order visibility, a sharp slowdown in enterprise deployment, or earnings reports that show AI capex outran end demand. Until that shows up in the numbers, we would treat this selloff as a reminder to be selective inside AI, not as proof that the whole theme just cracked. On that basis, the TickerSpark Score matters most where valuation and execution still line up, not where the market simply paid any price for the label.
Our take, not advice. This is opinion commentary — informational only, not personalized investment recommendations. Markets carry risk. Do your own research and consider your own situation before any trade.
▌The Daily Briefing · Free
A new stock idea, every evening.
One stock worth watching each weekday, plus the analysis behind it. Free, in your inbox.
