Market Trends Bearish 7

The ‘Citrini Selloff’: James van Geelen Warns of a Looming AI Capex Correction

· 3 min read · Verified by 2 sources ·
Share

Key Takeaways

  • Analyst James van Geelen, formerly of Citrini Research, has detailed a viral 'AI Doom Scenario' that predicts a massive market correction driven by the disconnect between AI infrastructure spending and actual enterprise ROI.
  • His thesis suggests that the current cycle of hyperscaler capex is unsustainable without a rapid acceleration in software-driven revenue.

Mentioned

James van Geelen person Bloomberg company Citrini Research product Odd Lots product

Key Intelligence

Key Facts

  1. 1The 'Citrini Selloff' thesis identifies a massive disconnect between AI infrastructure capex and enterprise software revenue.
  2. 2Hyperscalers are currently spending an estimated $200B+ annually on AI-related infrastructure, primarily GPUs.
  3. 3Van Geelen argues that much of the current AI revenue is 'circular,' driven by VC-funded startups buying compute from their own potential acquirers.
  4. 4The thesis warns of a 'fiber-optic style' bubble where overcapacity leads to a multi-year market correction.
  5. 5A key risk factor is the potential plateauing of 'scaling laws,' which would undermine the ROI for larger AI models.

Who's Affected

NVIDIA
companyNegative
AI Startups
companyNegative
Hyperscalers (MSFT, GOOGL)
companyNeutral
AI Infrastructure ROI Outlook

Analysis

The 'Citrini Selloff' has become a focal point for institutional investors and venture capitalists grappling with the sustainability of the generative AI boom. James van Geelen’s thesis, recently detailed on Bloomberg’s Odd Lots podcast, posits that the market is currently ignoring a fundamental gap between the hundreds of billions of dollars being poured into AI infrastructure and the relatively modest revenue being generated by AI software. This 'Capex Gap' is not merely a timing issue but a potential systemic risk to the valuations of both public tech giants and the private venture ecosystem.

At the heart of van Geelen’s argument is the 'circular economy' of AI. In this model, venture capital firms provide funding to AI startups, which then spend a significant portion of that capital on compute resources from hyperscalers like Microsoft, Google, and Amazon. These hyperscalers, in turn, report massive revenue growth and reinvest that capital into more GPUs from NVIDIA. Van Geelen warns that if the end-user—the enterprise customer—does not see a clear return on investment (ROI) from AI tools, this cycle will collapse. Once the VC-funded 'compute spend' dries up, the hyperscalers will be forced to slash their capex, leading to a sharp correction for the entire semiconductor and infrastructure supply chain.

For the venture capital community, the implications are particularly dire. Many 'AI-native' startups are currently valued on the assumption that they will eventually capture a massive slice of the enterprise software market. However, if van Geelen is correct, many of these companies are effectively 'compute-wrappers' that lack a sustainable moat. As enterprise budgets tighten and the 'hype' phase of AI transitions into a 'show-me-the-money' phase, these startups may find themselves unable to raise follow-on rounds, leading to a wave of shutdowns or fire sales. This scenario mirrors the fiber-optic bubble of the late 1990s, where massive over-investment in infrastructure preceded a decade-long period of consolidation.

What to Watch

Furthermore, van Geelen challenges the prevailing wisdom regarding 'scaling laws.' The belief that simply adding more data and more compute will lead to exponentially more capable models is the bedrock of current AI investment. If these laws hit a point of diminishing returns—as some recent research suggests—the economic justification for trillion-parameter models and multi-billion-dollar data centers evaporates. This would not only impact the 'Big Tech' players but would also invalidate the core thesis of many foundational model startups that have raised capital at multi-billion-dollar valuations.

Looking forward, the 'Citrini Selloff' serves as a cautionary framework for the next 12 to 18 months. Investors should closely monitor NVIDIA’s backlog and the earnings reports of major software-as-a-service (SaaS) providers. If enterprise adoption of AI agents and copilots fails to translate into significant productivity gains or new revenue streams by late 2026, the market may face the 'Minsky Moment' van Geelen describes—a sudden collapse in asset values following a long period of speculative growth. For VCs, the strategy may shift from 'growth at all costs' in AI to a more disciplined focus on unit economics and tangible enterprise integration.

How we covered this story

Every story in our startup coverage is assembled from multiple primary sources, cross-referenced for factual consistency, and scored along three independent dimensions: sentiment, operational impact, and source-cluster confidence. Single-source rumors and unverifiable claims do not pass our editorial gate. When a story shows "Verified by N sources" with N≥2, the development is independently corroborated; when N=1, we mark it explicitly so readers can weigh the signal accordingly.

Impact scoring uses a 1-10 scale weighted toward regulatory, financial, and operational consequence rather than coverage volume. A topic that runs in every outlet but moves no real decisions ranks lower than a niche regulatory filing that reshapes how operators in the startup space have to behave. Read our full methodology for the scoring rubric, our glossary for term definitions, and our trends index for the longitudinal view across the beat.