This one is going to be a little different from usual. I want to open with something personal and let the market piece follow. It has been an unusually hard week for me at home, and honestly, showing up to write this at all is part of the point. Let’s get into it.
1. The week the market did exactly what I have been writing about
If you have wandered through the Inspiration section of this blog, you will see some tough personal stretches in my lifetime. I relied on faith during those times and this week I was relying on that faith again.
My son has had some eye issues. He had surgery on one eye back in June and another on July 1st. The first surgery went well but an infection developed after the second surgery. The pain he experienced was excruciating which is tough to see as a Dad. His vision was and is at risk and there is nothing I can do but pray. That is what I do. You may not be a believer and that is completely fine and, honestly, I understand that belief in something you cannot see is challenging. I just know I might not be sitting here able to write without faith to take me through my toughest seasons.
Back to my son, he has dealt with so much in his life already, and watching your kid go through another hard stretch is the kind of thing that reorders every other item on your list. On top of that, I have had my own stress tied to personal health. I am not sure I ever mentioned it here but I required four weeks of daily radiation for some nasty skin cancer along with surgery for a different skin cancer. Lastly, which isn’t as stressful, I have been trying to get everything ready for both kids to head off to college in August, work has been crazy stressful, and the market decided to add its own noise to the pile.
Here is the part that surprised me. In a week when I could not afford to watch the market closely, the market did exactly what I have been writing about in this column for two months now. Iran flared back up. The U.S. reportedly struck targets. Oil moved. The Strait of Hormuz question came right back to the surface. Somewhere in the middle of the week, the Nasdaq had one very ugly day and then fought most of it back over the rest of the week. Chip stocks whipsawed both ways. New IPO. Meta jumped. The kind of week that could reasonably make someone anxious.
Instead of taking that path, I stepped away. I had to. My son needed me more than any headline did. And when I stepped back to check accounts at the end of the week, everything was fine. Not because I timed anything. Not because I made a brilliant call. Because I had already built a diversified portfolio, I had already decided I was in this for the long haul, and I had already accepted that no one can time the market. I owned companies I believed in for the long haul.
That is what discipline actually buys you. Not better returns in any single week. Not the ability to dodge every drawdown. What it buys you is the freedom to step away when your family needs you, and to know that the framework you built when things were calm will still be doing its job when things are not. The plan lets you not think about the plan.
It always seems impossible until it is done. Last week that quote was about a parkour athlete on one leg and a tiny nation pushing the World Cup champions to the brink. Some weeks it is about a father showing up to a keyboard when he has spent the past seven days praying for his kid. Some weeks it is about trusting a framework you built years ago and letting it hold while you focus on what matters more. All three count.
Hang tight. Dollar cost average when you can. Diversify. Trust the process. Especially in the weeks when you do not have the energy to do anything else.
2. Two big pieces of news from the week
Two stories worth naming this week. Meta made a significant move into a new area of AI that has real implications for the industry. And the second largest IPO of 2026 debuted on the Nasdaq. Both matter.
A quick word on terms for beginners
Before we get into the news, there are some terms that are going to show up in the next few paragraphs that you may not have heard before. If you already know these, skip ahead. If you are new to investing, here are user-friendly definitions.
- AI model — the actual software product that a company builds when we say a company “does AI.” ChatGPT is an AI model. Anthropic’s Claude is an AI model. Google’s Gemini is an AI model. A model is trained on huge amounts of data so it can produce useful outputs like text, images, or code.
- Agentic AI — the next step beyond a chatbot. An agentic AI is one that can take actions on your behalf rather than just answer questions. Instead of you asking it “how do I book a flight,” you tell it “book me a flight,” and it goes and does the booking. Agentic AI is a hot topic in the industry right now.
- AI infrastructure — the physical stuff needed to make AI work. Data centers full of specialized computer chips, the power to run them, the memory that feeds them, the cooling systems. It is very expensive to build. It is why the biggest tech companies are spending hundreds of billions of dollars a year on this.
- Capex (capital expenditures) — accountant word for money a company spends on physical assets it plans to use for years. Buying a factory is capex. Building a data center is capex. Meta’s $125 to $145 billion in “AI capex” for 2026 is money going into data centers and chips.
- HBM (High Bandwidth Memory) — a specialized kind of computer memory that is essential to AI. AI chips need to move enormous amounts of data very fast, and HBM is what makes that possible. Only three companies in the world make it in serious quantities — SK Hynix, Micron, and Samsung.
- ADR (American Depositary Receipt) — a way for a foreign company to have its shares traded on a U.S. stock exchange. When you buy an ADR, you are buying a certificate that represents an interest in the foreign company’s actual shares. For SK Hynix specifically, each ADR represents one-tenth of a common share. Practically, you can buy and sell it just like any other stock in your brokerage account.
- Hyperscaler — industry slang for the very largest cloud computing companies. Amazon (AWS), Microsoft (Azure), and Google (Google Cloud) are the classic three. They run enormous data centers and rent the compute out to other companies.
Meta had its best week since early 2024
Meta shares climbed roughly 6 percent on Friday and about 15 percent for the week. That is the stock’s best weekly performance since early 2024. The rally erased the stock’s losses for the year, meaning Meta went from being down for 2026 to modestly up in the span of five trading days. Three catalysts drove it. Caveat, I do own Meta in various accounts.
First, Meta released two new AI models from Meta Superintelligence Labs, which is now led by Alexandr Wang. Muse Image, a new image-generation model, launched Tuesday. Muse Spark 1.1, an upgraded agentic AI model, launched Thursday. Wang framed Muse Spark 1.1 as Meta’s strongest model yet for agentic and coding work. This is Meta stepping directly into the competitive space occupied by OpenAI, Anthropic, and Google. That is a meaningful shift. Meta has historically been more of a social media and advertising business. Now the company is putting focus on competing at the frontier of AI model development. Whether the models are good enough to actually compete is the question the market will answer over the next several quarters.
Second, Meta announced that its first in-house AI chip, code-named Iris, will begin manufacturing in September. Meta is targeting 14 gigawatts of computing power by 2027. The chip is being built with Broadcom and Taiwan Semiconductor. Building your own chip is a major structural move for a company that used to just buy silicon from Nvidia. It signals two things: Meta thinks it can save money on chips over time, and Meta thinks it needs to control its own AI infrastructure destiny.
Third, and probably most important, Bank of America published a note this week arguing that Meta is building its AI infrastructure at meaningfully lower cost than Wall Street had been modeling. That reframes the entire AI capex story I have been writing about for weeks. Meta guided to $125 to $145 billion in AI-related spending for 2026. If they are executing that spend more efficiently than the market assumed, then the capex burden is not the drag investors had been pricing in. That is what the 15 percent weekly rally is really about.
SK Hynix went public and it is bigger than most people realize
On Friday, South Korean memory chip giant SK Hynix debuted on the Nasdaq under the temporary ticker SKHYV (it converts to SKHY on Tuesday). The IPO priced at $149 per ADR (equivalent of a share price as defined above). The stock opened at $170, hit an intraday high of $177, and closed the day at $168.01, up roughly 13 percent from the IPO price. The listing raised $26.5 billion, which makes it the largest ever U.S. IPO by a foreign company, surpassing Alibaba’s 2014 record. Investor demand ran seven times oversubscribed. This is the second largest IPO of 2026 overall, behind only SpaceX in June which we discussed plenty.
Most retail investors do not know SK Hynix well because it has traded on the Korean stock exchange for years. I honestly didn’t either. Here is what matters. SK Hynix is the world leader in HBM (high bandwidth memory). Every time a hyperscaler buys more Nvidia GPUs to expand its AI compute capacity, SK Hynix sells more HBM to Nvidia. The company’s valuation has risen more than sevenfold in the past year on that dynamic. Its biggest customers include Nvidia and Apple. And per SK Hynix’s own SEC filing, the company controls 56.4 percent of the global HBM market.
This is the same memory shortage story I wrote about in mid-June when Apple and Microsoft raised prices on consumer electronics because memory costs had spiked. The industry called it RAMageddon. SK Hynix is one of the main reasons memory prices spiked. They and Micron and Samsung are the three companies that make almost all the world’s memory chips, and AI demand has soaked up their supply. That is exactly why SK Hynix chose to list on the Nasdaq now, at the moment when everyone in the AI supply chain is trying to secure long-term memory contracts.
The company is expected to be added to the Nasdaq 100 index in December 2026 during the routine rebalancing. That will trigger the same kind of forced buying from index funds that I explained in the SpaceX Nasdaq 100 post two weeks ago. If you own broad market index funds, you will own a slice of SK Hynix by year end whether you decide to buy it directly or not.
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 — July 4, 2026