This week the AI rally got bigger, the rules from last week kept proving themselves, and the Cerebras story I started telling last Saturday continued in a way worth watching. Three items, all connected. Let’s get into it.
1. The AI rally is much bigger than NVIDIA — here’s how it actually works
Most of the AI conversation centers on one company — NVIDIA. That makes sense. They design the chips that power almost everything happening in artificial intelligence right now. But if NVIDIA is the only AI stock you can name, you’re not seeing or don’t understand the full architecture driving this AI boom.
There’s an old expression from the 1849 California gold rush. Most of the prospectors who rushed in looking for gold never got rich. The people who got rich were the ones selling the picks and shovels — because every prospector, whether they struck gold or went home broke, needed picks and shovels. The shovel sellers got paid regardless of who won.
The same dynamic is playing out in AI right now. The AI labs — OpenAI, Anthropic (likely going public this year), all the names you hear about — are the prospectors. They’re competing in a fast-moving race where the eventual winners are far from settled. But every one of them needs the same physical infrastructure to operate. That infrastructure is the picks and shovels of the AI economy, and it goes a lot deeper than most people realize.
If you want one example that proves the point in numbers nobody can argue with, look at Micron. Micron makes the memory chips that AI servers cannot run without — AI workloads require a tremendous amount of memory and capacity to make these chips takes years to build. The result, this past week alone: Micron crossed $1 trillion in market value for the first time in its history. The Wall Street Journal reported that Micron’s climb from a $500 billion market cap to $1 trillion took just 48 days. The fastest such climb on record. Faster than NVIDIA. Faster than Apple. The fastest ever.
And here’s the part that is staggering: Micron stock is up roughly 928% over the past year. Not 28%. Nine hundred and twenty-eight percent. From $94 a share last May to $971 this Friday. Many analysts consider Micron one of the most overbought stocks after this week’s rally, which is a phrase worth holding onto for later in this post. We talked about FOMO last week. I had to heavily fight the FOMO urge myself this week although I would have made money.
Micron is the most dramatic example, but it is just one layer of the AI supply chain. Let me walk you through the rest of it, layer by layer — a tour of what it actually takes to run modern AI, starting at the chip and walking physically outward.
The chip itself. NVIDIA designs the GPUs that train and run AI models. ARM Holdings (ticker ARM) licenses the underlying chip architecture that many processors are built on top of — so even one layer beneath NVIDIA there’s a company getting paid.
The server the chip lives in. A GPU on its own is not useful. It has to sit inside a server, with the rest of the supporting hardware. Dell makes those servers — and as we’ll get to in a minute, Dell had one of the biggest stock weeks in its history because of exactly this.
The cables connecting everything. AI clusters move staggering amounts of data between servers, which requires high-speed fiber optic connections. Corning (ticker GLW) makes the glass fiber. Corning, the company that started in 1851. Suddenly an AI infrastructure stock. If you are unaware, Corning is also well known for its glass that is put into many things including Apple iphones.
The cooling systems. A modern AI server rack uses ten times the power a normal server rack drew just three years ago — enough that air cooling no longer works. Vertiv (ticker VRT) makes the liquid cooling systems that are now required.
The power management inside the building. Once electricity reaches the data center, it has to be distributed and managed safely at enormous scale. Eaton (ticker ETN) makes the equipment that does it.
The power generation feeding the building. The grid itself has to grow to support this. GE Vernova (ticker GEV) builds the turbines and grid equipment that generate and move the power.
And the backup power when the grid fails. A data center cannot go down. So every one of them has industrial-scale backup generators — and Caterpillar (ticker CAT), the heavy-equipment company founded in 1925, makes a lot of them.
Stop and notice what just happened. We started with a chip designer and ended at Caterpillar. A company most people associate with bulldozers and tractors is, in 2026, an AI play. Same with Corning, a 175-year-old glass company. The AI buildout is not just a software story — it is one of the most physically demanding industrial buildouts of the modern era. Somebody has to make the steel, the glass, the wire, the cooling, the turbines, the diesel generators.
That is the lesson I wanted to share this week. If you find yourself wishing you had owned NVIDIA five years ago, the more useful question is: which of the companies one or two layers out from NVIDIA still has room to grow as this buildout continues? The supply chain gets paid regardless of which AI lab wins. That is genuinely how I think about exposure to this trend — not as a single bet, but as a chain of suppliers. None of the companies I named are recommendations. They are examples of what each layer of the AI economy actually looks like, and a useful starting point for your own research if this is a space you want to understand better. There are many more companies in this space. Many!!
2. Dell vs. NVIDIA: the same lesson, opposite outcome — in seven days
Last week I wrote about NVIDIA’s earnings. Record revenue, guidance above expectations, every measurable thing was excellent — and the stock fell. The point I made was that the market prices things in. By the time NVIDIA reported, the good news was already in the price, so a beat was not enough.
This week, the exact same principle played out in reverse, and it is worth pausing on because you almost never get a clearer real-time illustration than this.
Dell reported earnings on Thursday evening. The numbers were not just good. They were so far above what Wall Street expected that the word “beat” barely covers it. Adjusted earnings per share came in at $4.86 against a consensus of about $3.00. Revenue was $43.8 billion against expectations near $35.8 billion. The company reported a record $51.3 billion backlog of AI server orders. None of that was in the price, because nobody had estimated numbers that big.
On Friday, the stock had its biggest one-day move in company history. Dell closed at $420.91, up 32.76% on the day. For the year, the stock is now up roughly 234%.
Stop and put the two weeks side by side. NVIDIA: strong, expected, priced in, stock fell. Dell: strong, unexpected, not priced in, stock exploded. Same principle, opposite outcomes, seven days apart. That is what “the market prices things in” actually means in practice — not that good news hurts a stock, but that the part of the news that was already expected is already reflected in the price. Only the surprise moves it. And on Thursday night, Dell delivered a surprise.
3. When everything is going up, discipline gets harder, not easier
The broader market just did something remarkable. The S&P 500 finished the week with its ninth consecutive weekly gain — the longest weekly winning streak the index has had since December 2023. More importantly, a nine week consecutive gain has only happened 10 times since 1945! AI infrastructure drove most of it. Dell up 234% on the year. Micron up 928%. NVIDIA still up double digits. Companies up and down the supply chain hitting new highs.
This week my personal FOMO lesson was Micron as discussed above. Last weeks example was Cerebras tied to their initial public offering. The stock ran up on pure hype, and people who chased it got hurt. The lesson wrote itself. Here’s where Cerebras stands one week later, for the update: the stock closed last Friday at $256.78, and this Friday it closed at 236.99 — 23.8% below its first-day close of $311 and well below the $386 high it hit when the hype peaked. The moves are getting smaller, which suggests the dust is starting to settle, but the settling point is meaningfully lower than where the excitement carried it. Lesson confirmed.
Micron is a harder lesson, and a more important one. Micron is not Cerebras. There is no hype problem here. The demand for memory chips is genuinely enormous, the company is genuinely well-positioned, the earnings are genuinely real. None of that is in dispute. But the stock is up 928% in a year. So the question for a regular person is not “is this a real company doing real things?” It clearly is a real company. The question is: how much of the next five years of real performance is already priced in today? When good news has been priced in by a factor of ten, even great news may not be enough to move the stock from here. That is the lesson NVIDIA just taught us two weeks ago.
When everything is going up, the urge to jump in becomes much stronger. That is when discipline gets harder, not easier. The same FOMO that pulled buyers into Cerebras at $350 a share is the FOMO pulling people toward anything green right now. And the temptation to chase Micron because the company is genuinely good is real. I mentioned my own internal debate but I stuck to rules I set. I may miss out. I am okay with that.
Rules built before the rally are the rules that protect you during it. The ones built during it are usually just rationalizations. The thing I said last week is still true. The moment you feel the FOMO is the worst possible moment to be making the decision. The decision should already have been made. None of that changes because the rally is bigger this week than it was last week. If anything, it matters more.
See you next Saturday.
Nothing here is personalized financial advice — just one person’s notes. The companies named are examples, not recommendations. Always do your own homework before you invest.
Weekly Notes — June 13, 2026