The Semiconductor Climax Run
Part 1 of 3 · Semiconductors just became 22% of the S&P 500 on the largest inflows ever recorded. A study of what happens at the end of a great trend — and how to read it.
Part 1 (Climax Run), Part 2 (Story is weaker than you think), Part 3 (Semi-Sell Rules)
«The market is designed to fool most of the people, most of the time.»
Jesse Livermore, American stock trader (1877–1940)
Something extraordinary is happening in the semiconductor sector, and almost nobody is framing it correctly.
The bulls see confirmation: record inflows, record leadership, the AI build-out validated by price. The bears see a bubble and have been wrong, painfully, for two years. Both are missing the more useful question — not is it a top, which nobody can answer, but where are we in the structure of a great trend, which the chart can actually tell you.
This is the first of three pieces on exactly that question. Today, in Part 1, we read the technical picture — what is happening, and the framework that makes sense of it. In Part 2, we turn the story over and ask whether the fundamentals underneath are as strong as the price implies. And in Part 3, we resolve the two-sided puzzle the way we resolve every position: not with a forecast, but with a plan — the sell rules that work whether this tops tomorrow or runs for two more years.
Let me show you what the tape is saying.
A Sector That Ate the Index
Start with the single most arresting fact in markets today.
Chart 1: The Philadelphia Semiconductor Index as a share of the S&P 500 — now 22%
Philly SOX as % of S&P 500 — 22%
For most of the 2000s and 2010s, semiconductors were a 3-to-5% corner of the S&P 500. A cyclical industry — important, but a corner. Today that corner is 22% of America’s largest companies. One narrow, deeply cyclical sector now carries more than a fifth of the broadest US benchmark.
The last time a single theme commanded this share of the index, it was technology in 1999 and energy in 2008 — and the parallel is worth sitting with, because of what came next. Technology peaked near a third of the S&P 500 in early 2000, then halved its index weight over the following two years and spent a decade in the wilderness. Energy reached roughly 16% of the index in 2008, then bled share for the next twelve years — the very collapse we wrote about in our energy piece. In both cases, the moment a single sector’s dominance felt most permanent was the moment it began to reverse. Concentration that extreme has, historically, marked something close to a peak — not a beginning.
Chart 2: Semiconductor ETF weekly flows — the largest on record
Semiconductor ETF weekly flows — record
And the money is still rushing in. The week of June 12, 2026, saw the single largest weekly inflow into semiconductor ETFs in the history of the data — larger than 2021, larger than the 2024 launch of the AI trade. Not early. Now, at the top of the move.
Chart 3: Momentum vs. minimum volatility — a five-standard-deviation overshoot
Momentum vs min-vol — 5-sigma overshoot
And the market has never paid up this aggressively for what is already working. The ratio of momentum stocks to low-volatility stocks now sits roughly five standard deviations above its long-term trend — more extreme than March 2000, more extreme than June 2008, more extreme than February 2021. Every prior reading near this level was followed by a painful unwind.
None of these three charts tells you to sell tomorrow. Parabolas run further and longer than anyone believes — what seems too high usually keeps going higher, and I have said exactly that for years. But together they describe something specific. They describe a late cycle. And late cycles do not require prediction. They require a framework.
The Anatomy of a Parabola
In the CAN SLIM tradition — William O’Neil’s framework, laid out in his classic «How to Make Money in Stocks», and refined by Mark Minervini — a great advance moves through three accelerating stages before it exhausts itself.
Chart 4: The SOXX semiconductor ETF, weekly — the three-stage parabola, with volume arriving last
SOXX weekly — three-stage parabola, volume circle
Look at the purple lines on the chart.
Speed Line One is the patient, multi-year uptrend off the 2018–2020 base. Price grinds higher, holding its moving averages, correcting and resuming. This is the part you want to own. Sit on your hands.
Speed Line Two is acceleration. The slope steepens; price pulls further and further above its own long-term averages. The crowd notices. The narrative hardens into consensus. It begins to feel less like investing and more like being carried.
Speed Line Three is the climax run — a sudden, near-vertical rise after a long advance. O’Neil studied the greatest winning stocks of the last century and found this final phase has a remarkably consistent fingerprint. It typically arrives only after a substantial prior advance — at least eighteen weeks from the original breakout, and often a 100%+ move already banked. Then the acceleration: a 25–50% surge in just one to three weeks, the stock up seven or eight days out of ten, printing the largest daily gain and the widest weekly spread of the entire move. Sometimes an exhaustion gap, where the day’s low opens above the prior day’s high. Sometimes «railroad tracks» — two consecutive wide, high-volume weeks that go almost nowhere. And often the corporate tells: a flurry of stock splits, breathless headlines, analyst target hikes, late-buyer FOMO.
And the signature, the tell, the thing that separates a climax from just another leg up:
Volume comes in last.
Look at the green circle on the chart. The heaviest volume of the entire multi-year run is arriving now, at the steepest part. This is the part most people get backwards. They think the surge of volume is confirmation — buy the strength. But at the end of a parabola, that volume is not the smart money arriving. It is the last money arriving. It is the dentist and the firefighter and the taxi driver finally deciding the thing is safe — precisely because it has gone up so much that it now feels safe. The institutions who accumulated quietly, stages ago, are the ones selling into that euphoria.
Here is the statistic that should stop every holder cold. O’Neil’s research on the biggest winners in market history found that stocks which top in a genuine climax run typically do not make a new high again for ten to twenty years — if ever. The climax is not a pause. For the great majority, it is the high-water mark of a generation. That is why the discipline is to sell into the strength, not wait for the breakdown: by the time the breakdown is obvious, the decade-long wait has already begun.
A record weekly inflow, into a 22%-of-the-index sector, five standard deviations stretched, during a near-vertical advance, with volume exploding. That is what topping action is made of.
This is not a new pattern. Livermore mapped it by hand a century ago — the accumulation, the markup, the climax, the test before the break. We will return to his speculative cycle in Part 3, because it does something remarkable: it tells you not just that a top is coming, but roughly where in the sequence we are right now. The instruments change. The psychology never does.
The Tape Says Tech — But Look Under the Hood
We never read price in isolation, so let us bring in our own screening — the engine we use to rank every sector on the «Good Story» (the fundamentals) and the «Good Chart» (the trend).
On the combined measure, Technology ranks first of all eleven sectors. Both halves light up: the story is strong, and the trend is the strongest in the market. On the face of it, that is a clean bull signal.
Chart 5: Sector Screening of arvy
11 key sectors screened by Good Story & Good Chart, Good Chart only / Forward Looking
But a number-one ranking is only as healthy as what sits underneath it — and underneath, Technology is not one thing. It is a barbell.
On one end: semiconductors and semiconductor equipment, going vertical, dragging the entire sector’s score upward — the parabola we have just been reading. On the other end: software, which is a different and far less happy story. We made the full case in «Software’s Good Story — and Why It’s Not Good Enough»: a sector in a primary downtrend, its moat under structural attack from large language models, its pricing power eroding, its multiple re-rating from 40x toward 20x and perhaps lower. We were invested in software for years. We sold.
So «Technology #1» is not the clean signal it appears to be. It is one euphoric, parabolic sub-group — the chips — masking a deteriorating one beneath. A sector whose leadership rests almost entirely on a single vertical theme is not a broad, healthy advance. It is a narrowing one. And narrowing leadership, where fewer and fewer names carry the whole index higher, is one of the oldest signatures of a late-stage market.
That is the «Good Chart» reading the same thing the parabola is: strength that is real, concentrated, and stretched.
In Fairness to the Bulls
Now let me argue the other side properly, because the bulls are not fools and this series will be worthless if it pretends they are.
The single most important fact about this trade is that the bears have been wrong, expensively, for two straight years — and not because they misread the chart, but because the underlying demand has been real. The AI build-out is not a story about nothing. The chips are sold out. The orders are real. The earnings the leaders are reporting are, today, genuinely there — not yet a mirage. And the businesses at the centre of it have authentic technological moats, measured in years of lead time, that no amount of capital can instantly replicate.
There is also a hard lesson in the tape itself: every investor who looked at this chart eighteen months ago, called it «too extended,» and stepped aside, has underperformed badly. «Too high» has been the single most expensive instinct in the market. A parabola, by its nature, spends most of its life looking unsustainable right up until the moment it becomes more unsustainable. Betting against a powerful trend simply because it is powerful is how careers end.
So we hold both truths at once. The setup is late-stage, concentrated and stretched — and the demand is real and the trend has humiliated everyone who fought it. That tension is not a contradiction to be resolved here. It is the entire reason this is a three-part study, and the reason Part 3 ends not with a verdict but with a plan.
Where This Leaves Us
So here is the picture at the end of Part 1.
A single cyclical sector has grown to 22% of the S&P 500. Money is pouring in at a record pace. The momentum trade is five standard deviations stretched. The chart shows a textbook three-stage parabola with volume exploding into the climax. And our own screening shows the sector’s leadership narrowing onto the chips alone, even as software — the other half of tech — quietly breaks down.
Every one of these is a hallmark of a late Stage 2, the climax phase of a great trend.
But — and this is the whole reason this is a three-part study and not a one-line sell call — late is not over. A climax can run for months. The most extended tape can become more extended. And there is a specific historical case, which we will come to in Part 3, where a move that looked exactly this exhausted reaccelerated into something even larger.
So we are not calling a top. We are doing something more useful: locating ourselves precisely in the structure of the trend, so that we are ready to act on whatever comes next — instead of surprised by it.
In Part 2, we turn the chart face-down and ask the harder question: is the story underneath this price as strong as the tape suggests? Because the most dangerous moment in markets is not when the chart looks stretched. It is when the chart looks stretched and the fundamentals are quietly hollowing out beneath it.
That is next.
«The market is designed to fool most of the people, most of the time.»
Livermore again. The tape is loud right now — louder than it has been in years. The question Part 2 will ask is whether it is telling the truth.
Thierry
Part 2 lands next: «The Story Is Weaker Than It Looks» — the fundamental cross-check on the AI-chip boom. Subscribe so you don’t miss it.
Legal Notice: At the time of publication of this article, the companies mentioned may or may not be portfolio holdings of arvy. Such securities may or may not be included in the portfolio at any point in the future. This document has been prepared solely for informational and marketing purposes and does not constitute an invitation, offer, or recommendation to acquire or sell any financial instruments or to engage in any other transactions. The relevant prospectus and key investor information documents (e.g. PRIIPs KID / Key Information Document) are available free of charge from arvy AG.
The people who were buying Nynix and Samsung to the top were the indivuals with leverage. There is nobody behind them. Now the Kospi is down 30% from top and nobody will buy. This is the typical case for a crash, independently of profits.