Hedge Fund Star Ken Griffin Thought AI Was "Garbage." Now He's Changing His Tune.

This reads less like a fundamental AI catalyst and more like a sentiment reset around what counts as monetizable AI. Near term, that is bearish for narrative-heavy names whose valuation depends on “AI” premiums rather than observable bookings or gross margin lift; speculative baskets like QBTS and RGTI are most exposed because their float is owned on story, not cash flow. NVDA is different: skepticism about AI-washing can actually improve capital discipline in the ecosystem and keep the market focused on the few vendors with real budget line-item exposure.

The second-order effect is that enterprise buyers may become more demanding on payback periods over the next 1-3 months, which can slow incremental software spend even if usage is rising. That is a headwind for any company pitching AI productivity without clean before/after metrics, and it raises the bar for application-layer software and services to show seat reduction, cycle-time compression, or margin expansion rather than usage anecdotes. If the next round of hyperscaler capex commentary or AI revenue disclosure softens, the entire complex could de-rate quickly.

The contrarian read is that this skepticism is healthy: it likely widens the gap between real AI beneficiaries and promotion stocks. Over 6-18 months, that should favor infrastructure and pick-and-shovel winners with measurable demand, while compressing multiples for pre-revenue “AI” and quantum names unless they prove repeatable commercial traction. What would falsify the bearish narrative on the hype names is a sequence of real bookings, not demos: sustained revenue acceleration, higher net retention, or a meaningful announced deployment pipeline.