The AI boom isn't a bubble — it's barely begun, Goldman Sachs says (www.businessinsider.com)

Goldman Sachs argues the AI boom is still in its opening act, not a bubble, and could generate trillions in long‑run value. In a note the bank estimates widespread generative AI adoption could add roughly $20 trillion to the U.S. economy (about $8 trillion flowing to companies) and lift economy‑wide labor productivity by ~15% over a 10‑year adoption window. Today’s AI spending—roughly $300 billion annually in 2025 and under 1% of U.S. GDP—is modest relative to past transformative waves (railroads, electrification, dot‑coms reached 2–5% of GDP), and Goldman says the productivity gains plus the massive compute and infrastructure needed justify continued investment. The analysts caution, however, that big spenders now aren’t guaranteed to be long‑term winners: hardware depreciates quickly, market structures are evolving, and history shows first movers often lose to later entrants who capitalize on overbuilt infrastructure. Vertical integration and semiconductor scarcity could preserve advantages for some, but multi‑model strategies and rapid tech change weaken incumbent moats. In short, Goldman sees a durable, compute‑heavy build phase ahead with eventual moderation as costs fall, but significant economic upside and strategic uncertainty about which firms will capture the largest share of value.