Original Viewpoints Summary

🤖 AI infrastructure and enterprise transformation

• AI/agent-native transformation consulting for large enterprises could become a bigger market than peak McKinsey / BCG / Accenture. The edge is not generic digitization anymore, but org redesign, management systems, and strategy built around AI-native workflows. That points to a potentially massive new services layer on top of the AI capex cycle. 2

🧠 Semis, valuation, and market setup

• The market is no longer asking $NVDA to merely beat. The bar has moved higher. Positioning now reflects expectations of a blowout plus strong forward framing, which raises the risk of a “good but not good enough” reaction into earnings. 8

• For memory names, dismissing PE with a lazy “cyclical stocks can’t be valued on PE” take does not hold up. If PE is rejected, an alternative framework such as PB / PS / DCF needs to be specified. The real point is not slogan-level labeling, but whether the chosen model matches the earnings cycle and industry structure. The pushback was already visible when Micron was at 400 last year. 14

• Semiconductor stocks have gone parabolic, and they are still the market’s sentiment hub. The Philadelphia Semiconductor Index (SOX) is up 50% over the last six weeks, making semis the key risk-on/risk-off battleground heading into Nvidia earnings. 13

🛡️ Portfolio positioning and capital allocation

• Soros Fund’s latest 13F reads like a defensive rebuild paired with an aggressive push into AI hardware. The setup is highly tactical—rotating through volatility to capture structural winners rather than just hugging broad market beta. The new BRK.B position of about $63.9 million also looks like a value anchor in a post-Buffett transition narrative. 1

• $COST and $WMT trading near all-time highs shows leadership is not exclusively AI-driven. But the catch is valuation stretch: both are trading at roughly 2x the forward multiple of the Mag 7 ex-Tesla, which makes the defensives look crowded and expensive even as they keep printing highs. 17

💻 Ownership, dilution, and long-duration compounding

• The Microsoft case is a brutal reminder of how much long-duration compounding can be lost through continuous selling. Bill Gates held about 45% at Microsoft’s 1986 IPO—some sources say 48%-49% intraday on listing day—then gradually sold down to below 1%, while the Gates Foundation has fully exited. Had those shares been held, the stake would be worth close to $1.5 trillion today. The underlying takeaway is that founder liquidity and philanthropy can massively reduce exposure to the very compounding engine they created. 18