Europe is priced like a museum, but the EUR/USD is holding strong—the market’s signaling something. If Europe can deliver on its investment renaissance over the next 5 years, it could surprise. The US is expensive—if you deploy LP money now, AI and robotics better deliver on GDP

Semiconductors is eating the world

Investors fear they missed the boat. But if you're asking that, you're watching the rearview. The right question is: what happens next? That’s how you board the boat if it’s already a 300-foot yacht while others are still staring at the dock.

👨‍🏫 A month ago, my father — a leading expert in Ancient Roman literature — turned 80. I asked what advice I should pass to the next generation. He gave me 3:

🌍 The world runs on data! 11,800 data centers power AI, cloud, & more. 🇺🇸 leads with 5,381, but China's Inner Mongolia site is the largest. 2025: US invests $100B+ in AI campuses. 🏜️ Middle East rises: Saudi's $10B push, UAE’s 5GW AI hub. Who’ll win the AI compute race? #AI

🚀 OpenAI created huge value, but open-source models make pure APIs tough to monetize. Like telcos, model providers (e.g., Anthropic) now compete with customers (see coding for example). Winners will master value capture via data moats, distribution, UX, & integration -not just

Crazy. 🤖🔐

Everyone says quantum is “decades away.” Yet Caltech just built a 6,100-qubit array. The commercialization timeline may not be linear. Progress is compounding faster than consensus admits. The bigger risk is often assuming the future looks slow. caltech.edu/about/news/cal…

Goodbye to Pasquale Pistorio, visionary leader, who built STMicroelectronics into a global powerhouse. A master of value creation in action. 🌟 en.ilsole24ore.com/art/goodbye-pa…

For years, semis were dismissed as too slow + too costly (“$5M+ for an ASIC” was the mantra entrepreneurs heard), while VCs chased apps, e-comm, fintech. Now the deeptech renaissance is here, driven by AI, geopolitics & supply chains. Big congrats to Rivos, chip developer on

Right idea, wrong decade. Market timing is the most expensive lesson in venture.

Marvell is reported to be in advanced talk to buy Celestial AI, a photonic startup with great demos but little revenue. The deal could be valued at $5B+. History says 80–90% of deals like this torch shareholder value. 3–5 year roadmap + AI FOMO = classic overpay risk. Great news

Amazon is the latest to release its own AI chip. Custom silicon is becoming a moat for hyperscalers (Amazon, Google, Microsoft), but only when paired with an equally strong software ecosystem. Hyperscalers can’t rely solely on Nvidia anymore because: 1. Nvidia captures 60–70%+ of

Amazing initiative

Okay Google, time to remake some good AI glasses. Please partner with EssilorLuxottica and don’t come up with the past nerdy design that only Silicon Valley engineers like.

Good start to the week. Paramount launched a $108b hostile bid for Warner Bros. Discovery 🍿 Inorganic growth in Hollywood won’t stop here. reuters.com/legal/transact…