Supply is still not catching up to demand in AI, says Solus' Dan Greenhaus

Key Concepts

• LLM Rotation: The phenomenon where users shift their preference and usage from one Large Language Model (LLM) provider (e.g., OpenAI) to a competitor (e.g., Anthropic).
• Compute Constraints: The ongoing limitation in the supply of high-performance computing power (GPUs/data centers) required to train and run AI models.
• Overbuilding: The economic risk of capital expenditure (CapEx) exceeding actual market demand, leading to future asset write-downs and valuation declines.
• Structural Remedies: Legal or regulatory interventions that force changes to a company's business model or structure (often discussed in antitrust contexts).

1. Market Dynamics and OpenAI’s Performance

The discussion centers on a recent Wall Street Journal report regarding OpenAI’s financial performance. The panel suggests that while OpenAI’s revenue figures may be under scrutiny, this does not necessarily signal a systemic collapse of the AI sector. Instead, it may reflect an "LLM rotation," where users are migrating to competitors like Anthropic.

• Key Argument: If OpenAI’s revenue is impaired, it could negatively impact the various startups and projects they have seeded. However, the demand for "compute" remains constant; if one provider loses market share, the computing resources are simply redirected to another provider.

2. The "Overbuilding" Thesis

A central debate is whether the massive investment in AI infrastructure will lead to a bubble burst due to overcapacity.

• The "Overbuild" Reality: The panel acknowledges that overbuilding is an inevitable outcome of the current AI arms race. CEOs in the sector have explicitly stated a preference for overbuilding rather than under-building to avoid missing the AI transition.
• Timing vs. Certainty: While a correction involving losses and write-downs is considered inevitable, the panel argues that the market is not there yet.
• Supply vs. Demand: The primary indicator for the health of the sector remains the gap between supply and demand. As long as demand for compute continues to exceed supply, the "overbuild" crisis remains a future concern rather than a present reality.

3. Earnings and Market Dependency

With major tech companies reporting earnings, the market’s current trajectory is heavily reliant on these results.

• Earnings as the Primary Driver: Despite geopolitical instability, the "earnings story" has been the primary catalyst for stock market growth this year.
• Expectations for Earnings Calls: The panel anticipates that upcoming earnings reports will confirm that supply constraints—not a lack of demand—remain the dominant narrative.
• Actionable Insight: Investors are advised to monitor whether companies report that they are still struggling to meet demand. If demand continues to outpace supply, the current bullish narrative for AI infrastructure remains intact.

4. Regulatory and Structural Outlook

Regarding potential regulatory pressure on OpenAI, the panel expressed skepticism about the likelihood of "structural remedies."

• Perspective: The consensus is that regulators are more likely to impose financial penalties or fees rather than forcing a breakup or fundamental restructuring of the company.

Synthesis and Conclusion

The current state of the AI market is characterized by a transition from initial hype to a focus on tangible earnings and operational capacity. While reports of OpenAI missing revenue targets have sparked concerns, the panel views this as a competitive shift (LLM rotation) rather than a decline in the broader AI utility. The overarching theme is that the industry is currently supply-constrained, and as long as demand for compute remains high, the risk of an "overbuild" bubble remains deferred. The upcoming earnings reports from major tech firms are viewed as the critical litmus test for whether the AI investment thesis remains sustainable.