Year follow-up on buying pandemic stock dip and AI reallocation

A retail investor's 4-year follow-up reports a 52% annualized money-weighted return from June 2022 through June 2026 on a portfolio that bought pandemic stocks after a 70%+ decline in 2022 and reallocated into AI stocks in 2023. Realized sales in 2026 returned -8.20% weighted by cost basis, while unrealized gains include Nvidia at +315.36%, Meta at +443.46%, and Intel at +206.24%. The investor sold most of CoreWeave in July 2026 and reallocated into memory stocks like SK Hynix.

This is my 4-year investment update following buying the dip on ‘pandemic stocks’ that declined 70%+ in 2022, then reallocating into AI stocks in 2023. I started sharing public updates once a year. Data in this update is as of June 2026. This will be a relatively short update since my thesis is relatively unchanged. See my past updates for more context: Recession opportunities original post /recession-opportunities/ 1 year follow-up on buying the dip on pandemic stocks /1-year-follow-up-on-buying-the-dip-on-pandemic-stocks/ 1.5 year follow-up on buying the dip on pandemic stocks /1-5-year-follow-up-on-buying-the-dip-on-pandemic-stocks/ 2 year follow-up on buying the dip on pandemic stocks /2-year-follow-up-on-buying-the-dip-on-pandemic-stocks/ 2.5 year follow-up on buying the dip on pandemic stocks /2-5-year-follow-up-on-buying-the-dip-on-pandemic-stocks/ 3 year follow-up on buying the dip on pandemic stocks /3-year-follow-up-on-buying-the-dip-on-pandemic-stocks/ Progress updates Below is the performance summary since my last update, a year ago in June 2025. Previous realized returns are unchanged and documented in the earlier updates. The indicates a short-term capital gain/loss investment held for less than one year . Realized sold in 2026 Apple AAPL : +16.77% Microsoft MSFT : +3.90% Netflix NFLX : -7.39% Palantir PLTR : -13.99% Shopify SHOP : -21.10% Super Micro Computer SMCI : -26.35% Tesla TSLA : +3.80% Weighted by cost basis, the combined realized return of -8.20%. These were the ones I had less confidence/conviction in. Unrealized current investments as of June 30 The holdings below are estimated using the reported cost basis and June 30 closing prices. The indicates a short-term capital gain/loss investment held for less than one year . Advanced Micro Devices AMD : +190.88% Alphabet/Google GOOGL : +17.60% Amazon AMZN : +13.20% ARM ARM : +208.19% ASML ASML : +47.41% Broadcom AVGO : +11.83% CoreWeave CRWV : -23.53% Intel INTC : +206.24% IREN IREN : +14.21% Meta META : +443.46% Micron MU : +183.97% Nebius NBIS : +154.87% Nvidia NVDA : +315.36% Oracle ORCL : -8.88% Taiwan Semiconductor TSM : +40.43% CoreWeave was the largest drag on the portfolio, and I sold most it in July and reallocated into memory mostly SK Hynix . Those numbers will be in the next update. Rate of return IRR My annualized money-weighted return for this portfolio from June 2022 through June 2026 is approximately 52% . That compares with 56% in last year’s update. Portfolio growth is net of deposits and withdrawals. The June 30 result is an estimate because I do not have a separate June month-end statement yet, but did my best to reconstruct it from statements, transactions and holdings. Visualized I used the performance/transaction to produce the visualization below. Portfolio growth is net of inflows/outflows. S&P is a price-return baseline. Investment thesis My investment thesis remains largely the same as it was in 2023. See a summary from this previous update /1-5-year-follow-up-on-buying-the-dip-on-pandemic-stocks/ . TL;DR: “…AI is another secular trend like PCs Windows, Mac , the internet browsers, search, social and mobile iOS, Android, wearables . The difference is that new technology like AI can now spread faster than ever before and get used in new ways. Every new epoch uniquely benefits from the past, potentially bending the growth curve in new ways. The other difference is that Nvidia has a monopoly position on the core technology driving this innovation. Therefore, the ~350% run up over the last 12 months doesn’t make Nvidia the stock of the last year, but rather it’s the stock of the next decade. The recent 3X gain will be a blip compared to what’s coming thanks to Nvidia’s CUDA moat , among other things .” 1.5 year follow-up on buying the dip on pandemic stocks – Nov. 2023 Past thoughts on AI Pragmatic agent-native architecture /building-agent-native-software/ HaaS will be the new SaaS /haas-will-be-the-new-saas/ Blind constraints, not blind spots /blind-constraints-not-blind-spots/ How AI makes for better software & companies /how-ai-makes-for-better-software-companies/ AI should write 50%+ of your code /ai-should-write-50-of-your-code/ Estimated tokens to merge ETM & other thoughts on AI coding /estimated-tokens-to-merge-etm-other-notes/ Why incumbents struggle with AI more than it seems /why-incumbents-struggle-with-ai-more-than-it-seems/ Consumer AI product retention /consumer-ai-product-retention/ Aggregating demand in the age of AI abundance /aggregation-abundance-ai/ Disruption is different with AI & more dangerous for incumbents /disruption-is-different-with-ai/ Build AI products, but don’t think of them like that /build-ai-products-but-dont-call-them-that/ Thoughts on AI so far /why-ai-is-everything-nothing/ Final thoughts I continue to be in the ‘ maybe I’m just lucky ‘ phase, and the major drop in IRR from ~70% to 52% shows just how fragile it is, especially doing it part time. All posts from this series: