Is the AI stock bubble actually popping now - from Broadcom miss to SpaceX overhang

[ParallelSelf99]

The timeline of the 2026 AI stock stress test: Broadcom left guidance flat on June 4 triggering a 4% Nasdaq drop. Quantum names lost 7-10% the same day. SpaceX liquidity rotation drained speculative capital through June 5-11. SPCX listed June 12 at $135 targeting $1.75 trillion. QNT fell below its $60 IPO price within 48 hours of listing. The S&P 500 is still up roughly 9% for the year but the 10-week winning streak has broken.

Ray Dalio has described the AI rally as being at 80% of dot-com peak euphoria. The Shiller P/E is at its second-highest level since 1871. The Bank of England and IMF both flagged AI valuations as stretched in Q4 2025. Palantir entered 2026 at 100x price-to-sales. The circular investment argument - hyperscalers spending hundreds of billions on AI capex that flows to chip companies whose valuations support the hyperscalers own stock prices - is getting louder

[StarLord67]

The difference between a correction and a bubble popping is whether the underlying earnings justify the valuations once the sentiment turns. Nvidia at $80 billion revenue is not Pets.com. Broadcom at $50 billion AI revenue is not Webvan. The multiples are stretched but the businesses are real

[Zach72]

Palantir at 100x price-to-sales entering 2026 was indefensible and has been repriced. Nvidia at its post-correction multiple is a very different conversation. The market is correctly differentiating between companies with real AI revenue and companies with AI narratives

[Freya]

Bubble or not, the Anthropic IPO at nearly a trillion dollar valuation and the OpenAI IPO both coming later in 2026 are going to be the defining sentiment tests. If both price well above their private valuations the bull case survives. If either disappoints badly the narrative breaks

[DarkEnergy]

Dalio at 80% of dot-com peak euphoria is a useful framing but it implies the remaining 20% could run for another 12 to 18 months before the break. Timing the top of a bubble using sentiment metrics is notoriously difficult and people who tried to short dot-com in 1998 got destroyed

[Sequence48]

The circular investment argument is the one I keep coming back to. If Meta, Microsoft and Google stop raising their AI capex guidance, Nvidia misses, AMD misses, Micron misses, TSMC misses. The whole structure depends on the hyperscalers continuing to spend at current rates indefinitely

[Sequence87]

The real test is AI revenue at the enterprise level. If companies are spending billions on AI tools and services and genuinely getting productivity returns, the capex cycle is self-sustaining. If the ROI does not materialise at scale in 2026 and 2027, the whole investment thesis gets questioned