Undervalued AI Tech

$MEANREV March 2026 Rebalance: Adding High-Conviction Mean Reversion Plays ▎ The portfolio has performed well at +84% all-time, driven primarily by our NVDA (25.6%) and GOOGL (17.9%) positions. However, both have appreciated significantly and no longer represent the "undervalued" side of our mean reversion thesis. We're trimming winners to fund three new positions that embody the core strategy: deeply oversold AI names with clear catalysts for reversion. ▎ Crypto positions (IBIT, ETHW, MSTR) slightly trimmed from 24% to 21%. Both BTC and ETH are down from our entry — we believe this is cyclical, not structural. Bitcoin's halving cycle historically drives 12-18 month recovery rallies, and Ethereum's deflationary supply mechanics post-merge remain intact. We're holding through the drawdown with high conviction in recovery but trimming modestly to fund new opportunities. ▎ New Positions: ▎ INTC — Intel (10%) ▎ Our highest-conviction mean reversion add. Intel is trading near decade lows at ~$22, down 65%+ from 2024 highs. The market is pricing in a failed turnaround, but the fundamentals tell a different story. New CEO Lip-Bu Tan — a legendary semiconductor investor — is executing a credible transformation: the 18A process node is on track, Intel Foundry Services is landing its first external customers, and the CHIPS Act is providing $8.5B in direct funding for US fab construction. At sub-1x price-to-sales with $50B+ in annual revenue, Intel is priced for permanent decline while showing early signs of operational inflection. If the foundry strategy gains traction, there's 2-3x upside from here. This is exactly the kind of asymmetric setup our strategy was built for. ▎ CRWD — CrowdStrike (8%) ▎ CrowdStrike has sold off ~35% from its mid-2024 highs after the July outage incident knocked out 8.5 million Windows machines worldwide. The selloff was justified in the immediate aftermath, but six months later, the business has proven remarkably resilient: gross retention remains at 97%, net new ARR reaccelerated, and the company posted record free cash flow in Q3. The outage was a configuration deployment error — a one-time human process failure, not a product architecture flaw. CrowdStrike's Falcon platform remains the gold standard in endpoint security, and every enterprise needs AI-native threat detection. At ~15x forward revenue (vs 25x+ pre-incident), the market is still pricing in lasting damage that the financials aren't showing. Classic overreaction → reversion pattern. ▎ SNOW — Snowflake (7%) ▎ Down 55% from its all-time high and largely abandoned by growth investors who rotated into pure-play AI names. But Snowflake sits at a critical intersection: every enterprise AI deployment needs a data layer, and Snowflake's Data Cloud is the leading platform for AI-ready data warehousing. New CEO Sridhar Ramaswamy (former head of Google's AI division) is aggressively pivoting the product toward AI-native workloads — Cortex AI, Document AI, and Snowpark ML are all seeing rapid adoption. Revenue is still growing 25%+ YoY at $3.5B+ run rate. At ~12x forward revenue (vs 30x at peak and 20x industry avg for high-growth cloud), the valuation reflects deep pessimism about growth durability. If AI data workloads drive even modest re-acceleration, the reversion potential is 50-80%. ▎ Trimmed Positions: ▎ NVDA trimmed from 25.6% to 18%. Nvidia remains the undisputed AI compute leader, but at 30x+ forward earnings, it's no longer undervalued by any traditional metric. We're taking profits on our largest winner to reallocate toward names with more reversion upside. Still our biggest single position — we're trimming, not exiting. ▎ GOOGL trimmed from 17.9% to 12%. Alphabet is still undervalued relative to its AI capabilities (Gemini 2.0, Google Cloud growing 30%+, YouTube AI monetization), but we're reducing to fund higher-conviction mean reversion plays. At 20x earnings, GOOGL has less upside asymmetry than our new adds. ▎ META and MSFT trimmed to 5% each. Both are quality holds but are fairly valued — they're portfolio stabilizers, not mean reversion candidates. ▎ Portfolio Thesis: We're rebalancing from a "hold the winners" posture back to the active mean reversion strategy that generated our +84% return. The new adds (INTC, CRWD, SNOW) are all trading 35-65% below intrinsic value with identifiable catalysts. Crypto remains a meaningful 21% allocation reflecting our conviction in cyclical recovery. The portfolio is now balanced between asymmetric reversion plays (25%), core AI infrastructure (32%), mega-cap AI (22%), and crypto (21%).

Rotating out of our Amazon & Apple positions: taking profit and adding a NEW Pure Storage position. $PSTG is a high-conviction AI infrastructure bet: explosive data growth is driving demand for high-performance, energy-efficient flash storage, while Pure’s subscription model creates sticky, recurring revenue and expanding margins. A key catalyst was seeing Gavin Baker of Atreides Management and major AI leader) add it as a meaningful position—reinforcing the view that Pure is emerging as a long-duration compounder at the core of modern cloud and AI workloads.

Increasing our crypto exposure. Crypto is increasingly becoming a key financial rail with this recent crypto wave being deemed the “institutional cycle”, where real financial institutions are adopting it. I also believe crypto will be a key method to verify the authenticity of content with AI’s rapid growth. Ethereum is at a monthly low and Bitcoin is going through a pullback, so I view this as a good entry point to establish and increase our positions. Still long NVDA but just decreasing our exposure to make room.