Stock Market Weekly Retrospect
August 2026
If you're invested in the AI theme, you probably felt the drop in July - even though plenty of people missed it entirely. But even after a rough month, the market never runs short of surprises and interesting stories.
Earnings Season
I would understand bears if AI infra companies had no revenue, and were talking about dreams in 2035. But they’re not.
Earnings after earnings, companies delivered. A lot of them sold off, not because of the disappointing earnings, but because they simply ran too much, or analysts set unrealistic expectations.
One of the hottest earnings on August 6, 2026 was definitely $AAOI. They delivered. The story got stronger, not weaker. They can’t produce fast enough. AOI makes optical transceivers, the components that let AI data centers move data at the speeds these clusters need, and right now demand is outrunning their ability to build them. It’s about the ramp and mid 2027 transceiver output, and all earnings from now till then are temperature checks on whether that ramp stays on track.
Nebius is another name that sent a message loud and clear: AI is not dead.
This is almost a 100% increase in ACV (Annual Contract Value, the revenue customers have committed to), and it tells you a lot about how strong current demand really is.
Revenue $582.3M vs. $573.9M est.
Adj. EBITDA $236.2M vs. $172.6M est.
ARR $3B (+58% QoQ)
2026 outlook:
Year-end ARR of ~$7-9B (reaffirmed)
Raised contracted power from >4GW to 5GW
The bears
Fintwit’s favorite Michael Burry allegedly shorted both Micron and Nebius, and even after Nebius’s incredible run, there’s a message circling around how he doubled down. Now, since he isn’t obliged to disclose his positions, I’m not convinced he actually shorted anything. This could be a move to get more PR, but so far the PR has been mostly AI made clown faces of Burry.
I truly believe he is a smart guy, but he picked one of the worst companies to short. Their financials have never looked better, and sentiment just improved once again. Shorting in July, I’d understand, but in August, you have to be either crazy, brave, or both.
Memory is dead, it’s optics’ time to shine!
Here is what started this “trend”.
Now, there’s a lot to unpack, and a lot of it was said in the comments. Jukan seems to have changed his position on memory, and the community reaction was strong.
I made a satirical post on the same day.
A lot of memory bulls and Fintwit creators turned bearish on memory (short term, they say), and cited optics as the short-term winner instead.
My message was clear: no one knows what will fly next. We can guess, but we can’t claim to know. Memory recovered well last week. Optics (apart from AAOI) didn’t do so great.
While I disagree with Jukan, I very much condemn all the insults and name-calling sent his way. People should be allowed to share opinions without being lynched.
Then came this tweet, which got retweeted hundreds of times within an hour.
It was another sign of retail sentiment flip-flopping on memory, even though nothing about the fundamentals actually changed for the worse in July.
Citadel…
I ran into an excellent article recapping July, and giving important information, stats, and updates regarding overall market performance.
Some highlights.
”While volumes normalized from the extremes of the prior two months, retail investor behavior changed materially over the course of July. As many of retail’s highest-conviction AI positions continued to weaken, investors increasingly shifted from buying weakness to reducing exposure. That transition accelerated into the final week of the month, culminating in what is now on pace to be the largest week of retail equity selling since 2022.”
This is one of the reasons most retail investors lose money. They buy the top, and sell the bottom. A lot of companies, including our dear Citadel, “helped” retail sell more by leaking false stories about rate hikes.
Now onto Earnings Season performance (I spoke about it at the beginning)
What bubble? Earnings are performing excellently, yet stocks are still being sold off. Not saying it’s a bad sign, but it was showing increased volatility in July.
“Stronger fundamentals have not been accompanied by multiple expansion. Instead, stronger earnings have coincided with a meaningful compression in valuations.”
Sounds familiar? Most noticeable with memory names.
In any case, we went through the most important tests: the FOMC decision and hyperscaler capex updates.
I loved the last sentence from Citadel’s report, as it said what I was saying
“The technical reset we have been waiting for has largely occurred. July did not change the structural bull market. It reset it.”
July was a bigger breather than most expected but the one we needed. Stocks can’t endlessly go up.
U.S. Urges Apple Not to Buy Chinese Memory Chips
This ties directly into the memory supply crunch we’ve been talking about all month. Howard Lutnick said the Trump admin opposes Apple buying memory chips from China, telling Apple this "plainly" after touring an Apple facility in Houston.
Apple is reportedly testing chips from CXMT and Yangtze Memory Technologies (YMTC) as it faces a severe memory shortage driven by AI data center demand, which has driven up prices and forced consumer electronics makers to raise prices.
Apple can legally buy off-the-shelf parts from the Chinese suppliers without a license, though customized chip work would require U.S. government approval. Apple COO Sabih Khan wouldn't confirm the Chinese chip testing but said the company has to "look at all options," noting memory isn't heavily customized compared to other iPhone components.
Low VIX
One thing that doesn’t get talked about enough right now: the VIX is sitting low, and that’s not the all-clear signal people treat it as. Low volatility doesn’t mean the market is calm, it usually means it’s compressed, and compressed things tend to snap rather than drift. Historically, some of the sharpest volatility spikes have come right after extended stretches of a quiet VIX, precisely because everyone gets comfortable at the same time and positions accordingly: short vol, max leverage, light hedging.
This isn’t a call for a crash. It’s a reminder that a low VIX measures recent calm, not future calm. After a month like July, with earnings, capex updates, and Fed decisions all still in play, this is not the time to relax.
Do let me know if you enjoy these recaps!
Great post, thanks.
Noone really should be writing about memory without understanding what cHBM, HBF and HBM5 are and what they mean.
The most rational reason to short Micron is the fact they are losing the technology race. Moats are built on a technological lead. The tech leaders on cHBM: Hynix and Samsung. On HBF: Hynix and Sandisk. On HBM5: Hynix and Samsung. Which name is missing on all three? Micron, the everlasting follower. (And any and all Chinese companies, of course.)
cHBM will mean the tech leaders in 2028 will lock in customers long term, giving the first to offer cHBM (likely Hynix, but at least possibly Samsung) incredible pricing power for a very long time.