‘Vibe revenue’: AI companies admit they’re worried about a bubble (www.cnbc.com)

At the Web Summit in Lisbon, CEOs from AI firms including DeepL and Picsart warned that soaring valuations look “exaggerated” and may signal a financial bubble, echoing concerns from investors such as Goldman Sachs, Morgan Stanley and activist Michael Burry (who has accused hyperscalers of understating chip depreciation and placed bearish bets on Nvidia and Palantir). Executives coined the term “vibe revenue” to describe startups winning big valuations with little real sales, even as leaders like Lyft’s CEO and Cohere’s CFO stressed that AI’s industrial potential remains transformational and enterprise demand should stay strong into 2026. The debate matters because it cuts to resource allocation and the economics of scaling AI: Accel forecasts 117 GW of new AI data‑centre capacity by 2030 — roughly $4 trillion in capex over five years, requiring about $3.1 trillion of revenue to pay back — driving massive investment in chips, power and model training. Yet some investors argue those figures and energy needs may be overstated, and enterprises are still struggling with adoption and integration of “agentic” tools that act autonomously on users’ behalf. The net implication: while long‑term demand and model-driven infrastructure appear robust, market prices, capital intensity and profit accounting are under fresh scrutiny — increasing the odds of a correction even as AI products continue to advance.