Sleep Well Portfolio (July 2026) - Recovery continues, AI's ROI questioned.
AI-insulated positioned rallied. Five losers out of 13 holdings; portfolio recovery continues.
The Sleep Well Portfolio consists of time-tested, market-leading companies. We screen them with a rigorous checklist and regularly track their theses to determine when to buy. So far, we have made one mistake, but are happy with the current 13 holdings. All about us here.
Hi all,
In July, we saw more uncertainties and speculations around AI spending and return on investment (ROI). Micron (MU), Nebius (NBIS), and other AI infrastructure names fell as much as 50% from the recent peak. Mr. market was highly unstable. We, on the other hand, had a quiet month. The portfolio recovery continued, but not much happened. August and September will be more eventful months to report as the Q2 reports roll in, so I’ll keep this short.
AI’s ROI not clear
There is no doubt AI will revolutionize everything digitally in the next 5 years. Elon Musk goes further, predicting AI will also dominate the physical world by 2036, driving the greatest economic transformation. Combining both means we will enter a ‘quasi-infinite economy’, where digital intelligence with vast numbers of physical robots creates a framework where goods and services become so abundant that money and traditional jobs may become optional by 2036.
If you share Elon’s vision, I guess AI’s ROI will be substantial for all.
For now, even as hyperscalers continue to expand backlog (AWS expects to reach $500B by 2027), I can’t say ROI is clear.
First, these hyperscalers are also expanding their CAPEX budget at a rapid clip (Amazon increased its capex budget to $220B for 2026). Then, we also saw SpaceX and Meta early in July announce they were renting out excess capacity (on a 90-day basis), and thirdly, Ramp’s latest AI Index shows the median firm spends $11.38 per employee per month on AI, which is ~700x less than the top 1% of firms spending ~$7,500 per employee per month.
So, if there is ROI, AI usage and monetization are only tightly concentrated to a small group of power users doing important work.
For the majority, AI usage and ROI are still unclear, in my view.
That said, I remain closely watching Amazon, Microsoft, and the likes as an investment directly in AI. Certainly, the constraints in chips (China) and power and cooling (US) are very interesting to explore. I wish to share some of these names in the future.
Portfolio recovery continues
Despite the volatility and liquidity pull from AI, the portfolio’s recovery has kept the pace from June. It is happening faster than I thought, as it’s been just 6 months since the bottom of the software apocalypse, which was also when we evaluated the AI risk for all holdings (link here). The takeaway is that whenever the worst headlines are on the news, that’s an indication of the bottom. Since then, most of the software holdings have shown no signs of disruption, and shares have rebounded 20-30% (Veeva, Fortinet, CrowdStrike, Kinsale, Cellebrite). Our ecommerce and fintech (Sea Limited, Mercado Libre, Grab) are lagging, but I remain very optimistic about their future.
If I hadn’t had a large allocation for these names, I would be adding more already. History shows opportunities may come. The last drawdown of this scale lasted a year and a half after the war in Ukraine started. And, learning from the last period, all in all, we made 10 purchases and 1 sell. Almost all have been highly accretive to the portfolio so far. I believe patience will win.
Before we review the portfolio. I have a small change to make. I find the monthly Best Buys, Thesis Tracking, and Portfolio updates counterproductive to long-term investing. They encourage overthinking and tinkering in such a short period that makes very little difference to the business. So going forward, I will focus on the quarterly updates (Thesis Tracking), keep the monthly portfolio update brief, and transition to a longer-form quarterly portfolio update. My primary goal remains building a sleep well portfolio for my daughters, and the frequency of documentation makes it less productive, in my view.
Summary of July 2026
July was a non-event month after an incredible resurgence of software in June. Intuitive Surgical reported Q2; nothing changed in my view, despite a 10% drop, so the in-depth writeup on why I bought below the company remains relevant.
Deep dive here in case you want to know everything about the company.
Here is how the portfolio looks today:
After 8 months of me ‘looking wrong’, the last three months are starting to prove that the hardest thing in investing is actually the waiting, and even harder is slowly buying when everyone is selling. No one quite knows when the bottom hits, but you have to know what is a reasonable deal for you personally, i.e., what is your financial goal, what is enough, what is the risk, how long can you wait, how confident are you about the value of the business you are buying, etc.
As always, time will tell. I have survived the 2008, 2015, 2020, 2022, and 2025 down years. Market cycles are a feature of long-term investing, not a bug.
By 2037 (when my daughters will have full control of the Sleep Well Portfolio), none of this would matter. I continue to put aside some savings for them and own more winners - ‘winner keeps winning ’-in both bear and bull markets.
If you are new to SWI, check out our FAQ and Owner Manual.
If you are already an annual member, you can access the Sleep Well Portfolio and Thesis Tracker via the link below.
