Listen to the audio version of this article (generated by AI).
The machines couldn’t keep up.
By mid-afternoon, the ticker tape in lower Manhattan was running more than two hours behind the actual selling. Brokers were quoting prices that were already outdated, and, in a single session, the Dow Jones shed nearly 6%. It was the worst one-day drop since the Crash of 1929. The next day, the morning papers declared the postwar bull market dead.
They called it Blue Monday, the depths of the “Kennedy Slide” of 1962. The S&P 500 fell roughly 27% from its highs. Yet, within 14 months, the market had clawed back every lost point. The boom that supposedly died that spring went on to run for several more years.
Sound familiar?
The past week’s been peppered with headlines like:
The Philadelphia Semiconductor Index (SOX) dropped 25% from its highs – the worst selloff in AI infrastructure stocks since ChatGPT launched in late 2022, once you set aside Liberation Day as the black swan it was.
Individual stocks dropped 20%, 40%, even more. Painful. Awful. I haven’t enjoyed a single minute of it.
But I’m also not terribly worried. Because, when I recorded this podcast on Thursday morning, the tape sent a signal most investors missed: while semis cratered 25%, the S&P dropped just 2%. When the dot-com boom actually ended in 2000, the damage spread like a contagion across the entire market. Today’s selloff looks far more like late 1995 or late 1997 – violent recalibrations mid-boom, with years of upside still ahead.
Meanwhile, the top of the spending funnel keeps widening.
Alphabet (GOOGL), Tesla (TSLA), and Taiwan Semiconductor Manufacturing (TSM) all just raised capex forecasts. Nvidia (NVDA) is backstopping $250 billion for a 10-gigawatt OpenAI data center. Morgan Stanley (MS) now models hyperscaler spending at $1.2 trillion for 2027.
So why are we not backing up the truck yet? Because one technical level – a line in the sand on the semiconductor charts – will tell us whether this is a recalibration or a genuine ending. In this episode of Being Exponential With Luke Lango, our host, Brook Teferi, talks with me about exactly where that line sits, what a convincing bounce looks like, and the handful of sectors I’d pivot toward if it breaks.
Watch the full episode below:
The Bear Case to Watch
I assure you, the worst AI selloff since ChatGPT launched in late 2022 is not, by itself, evidence the AI boom is ending. It feels like the end, sure. But feelings make terrible trading decisions.
Let’s not sugarcoat it. What the markets just lived through is real pain. Real capital destroyed. Nobody at Hypergrowth Investing enjoyed a single day of it.
But let me be clear: painful and dangerous are not the same. And the data underneath the pain tells a very different story than the headlines do.
You could think of it this way: markets move two steps forward and one step back. Sometimes, after ten strong steps, you get four or five steps back all at once. The trend, far from reversing, is simply catching its breath.
At the end of the day, the capital keeps flowing, and the flow is accelerating, not slowing.
The bear case worth taking seriously is that the hyperscalers have drained their cash flows funding this buildout, and incremental spending increases now have to come from debt markets rather than existing cash. That introduces leverage into a trade that used to run on cash. Add in rising long-end interest rates and a growing populist backlash against AI infrastructure, and you get the ingredients for a genuine slowdown in end-market demand.
That thesis is directionally correct, but the timing is early. The real risk window is more likely around 2028 to 2029, tied to the next election cycle and the political appetite for AI regulation, not to anything happening in the market today.
Right now, demand shows no sign of cracking. Cloud growth numbers keep beating estimates. Companies keep shipping new models. Uncle Sam, for his part, has made AI infrastructure a national priority, and that kind of political backstop does not evaporate overnight.
The One Chart You Actually Need to Watch
Here is where it gets tactical, and here is the part I recommend you do not skip when you watch the full episode. We will not back up the truck, meaning we will not add aggressively to AI positions, until the semiconductor ETFs SOXX and SMH prove themselves at a specific technical level: the 100-day moving average.
Every local bottom in SMH since this boom began, in April 2023, April 2024, November 2025, and March 2026, has held right around that 100-day line, or within about 10% below it.
If the current selloff bottoms in that same range and produces a convincing bounce, 5-10% higher, retaking resistance rather than getting rejected by it, that is the market confirming this is a recalibration, not an ending.
If it breaks that range instead, that is the market telling you this time is genuinely different, and you do not fight the tape.
We walk through exactly what a “convincing bounce” looks like on the charts, plus which sectors, from Chinese internet stocks to defense to energy, could take the baton if AI stumbles further before it does not. That full breakdown, chart by chart, lives in .
But private markets run on a different clock entirely. The capital pouring into AI infrastructure right now isn’t waiting for SOXX to find its 100-day moving average. It committed months ago, and one hyperscaler’s earnings call isn’t enough to spook the demand signal.
That lag between public and private markets is your opportunity.
Public investors are stuck watching a chart for confirmation. Private investors already have their answer.
That’s the case I just made live at , including the free recommendation I handed attendees on the air. The replay is still up, but not for long, and who knows how long the recommendation will remain open.
