Will the AI Bubble Burst in 2026?

AI stocks currently trade at 50–100x forward earnings, levels reminiscent of the dot-com era peak in 1999–2000. NVIDIA's price-to-earnings ratio has spent extended periods above 70x even as revenue scales rapidly. Broader AI-adjacent software companies — Palantir, C3.ai, UiPath — trade at 20–50x forward revenue in some cases. When market sentiment shifts, assets priced for perfection correct violently. Historical precedent from the dot-com crash, the 2022 growth stock rout (Nasdaq -33%), and the 2021 SPAC collapse all demonstrate how quickly euphoria-driven multiples compress under rising rates or missed earnings expectations.

Unlike the dot-com bubble, where many companies had no revenue model whatsoever, major AI infrastructure companies are generating massive and rapidly growing revenue. NVIDIA crossed $100B annualized revenue in 2024 — a genuine business milestone. Microsoft Azure AI services, Google Cloud AI, and Amazon Bedrock are all growing 30–50% year-over-year. This fundamental revenue base provides a meaningful floor that did not exist for Pets.com or Webvan. The bull case argues that current valuations are justified by a multi-decade AI adoption curve, where total addressable market expands far beyond current enterprise software spend.

Hyperscalers — Microsoft, Google, Amazon, Meta — collectively committed over $200 billion in AI infrastructure capex for 2024–2025, with plans to maintain or accelerate this through 2026. If AI application revenue growth disappoints, this capital investment creates a significant overhang: depreciation charges hit earnings, data center excess capacity drives price competition, and GPU demand softtens. Analyst estimates suggest hyperscalers need to generate $600B+ in AI-driven revenue to justify their current capex cycles. If that threshold is not reached by 2026–2027, earnings revisions could trigger a sector-wide re-rating.

US export restrictions on advanced AI chips — particularly NVIDIA's H100 and H200 GPUs — to China represent a structural risk to NVIDIA's revenue concentration and to global AI supply chains. The US-China semiconductor rivalry has escalated through successive rounds of export controls (October 2022, October 2023, and ongoing updates). A sudden tightening that cuts off Middle East or additional Asian markets, or a retaliatory Chinese restriction on rare earth minerals critical to semiconductor manufacturing, could materially disrupt AI hardware supply chains and rattle investor confidence simultaneously.

“Grantham, who correctly called the dot-com bubble, the 2008 housing crisis, and the 2021 growth stock peak, has argued publicly that AI valuations exhibit the euphoria-driven excess characteristic of historic bubbles. He notes that 'truly transformative technologies — railroads, electricity, the internet — have historically generated spectacular bubbles even when the technology itself succeeded.' His view: AI will transform the world and the bubble will still burst. He places AI stocks among the most overvalued assets in market history by normalized earnings metrics.”

来源: Jeremy Grantham, GMO Co-Founder

“Sequoia partner David Cahn published a widely circulated analysis estimating that AI infrastructure spending implies a need for $600B in annual AI application revenue to justify current investment levels — revenue that does not yet exist. The piece, titled 'AI's $600B Question,' was updated from an earlier $200B estimate as capex announcements multiplied. Cahn stopped short of calling a bubble but described the current capex-to-revenue ratio as unsustainable in the medium term, arguing the market is 'pricing in a future that may be 5-10 years away.'”

来源: Sequoia Capital (David Cahn)

“Wood maintains that AI represents the largest technological convergence in history — combining robotics, energy storage, AI, blockchain, and multi-omics biology — and that current valuations understate long-term potential. ARK's models project AI-driven productivity gains adding $200 trillion in global economic value by 2030. Critics note that ARK's flagship fund fell 75% from its 2021 peak before recovering partially, and that Wood was also bullish on tech through the entire 2022 growth stock correction. Her track record on timing is mixed despite directional accuracy on AI importance.”

来源: Cathie Wood, ARK Invest CEO