Are we dismissing AI spend before the 6x lands?

Summary

The article discusses the upcoming surge in AI compute capacity, driven by advancements in chip technology. It highlights the potential implications of this growth, while also acknowledging challenges like power constraints and training lags. The author expresses excitement about the future capabilities of AI models.

Key Takeaways

1. AI silicon flow is increasing dramatically - and really starts gathering steam into 2026.
2. Between ChatGPT first launching and the end of 2026, the world will have nearly 50x more compute installed and available for us.
3. Opus 4.5 and Gemini 3 were likely trained on the 2024 install base.

Statement Breakdown

• Claimed Facts: 60% of statements the article presents as facts
• Opinions: 30% of statements classified as editorial or subjective
• Claims: 10% of statements surfaced for additional reader evaluation

Credibility & Bias Reasoning

Credibility assessment: The article presents data from Morgan Stanley and the author's own calculations, but also acknowledges the guesswork involved. The author provides context and limitations to their analysis. The article relies on industry knowledge and data, enhancing its credibility.

Bias assessment: Technological Optimism. The article expresses excitement and anticipation regarding the future impact of AI compute. It focuses on the potential benefits and advancements driven by increased AI infrastructure. While acknowledging potential challenges, the overall tone is optimistic about the transformative power of AI.

Note: While the article presents data and analysis, some claims rely on estimations and industry rumors. Consider the author's optimistic perspective when interpreting the potential impact of AI advancements.

Credibility flag: Cautious Optimism

Claimed Facts (6)

• This is a statement of fact regarding Morgan Stanley's research.
• This is a factual statement based on the provided data.
• This is a factual statement about Broadcom and AMD's capacity.
• This is a statement of fact based on industry rumors.
• This is a statement of fact based on Sam Altman's interview.
• This is a factual statement based on METR's work.

Opinions (6)

• This is an opinion about the critics' evaluation of AI models.
• This is an opinion about the limitations of the data.
• This is an opinion about the significance of the growth.
• This is an opinion about the limiting factor of the growth.
• This is an opinion about the support for the author's observations.
• This is an opinion based on personal experience.

Claims (5)

• The claim of a '6x wave' is presented without concrete evidence, relying on future projections.
• The term "napkin math" suggests a lack of rigor and precision in the calculations.
• The assumption of an "aggressive ramp" is speculative and lacks solid backing.
• The analogy to railway construction is a broad comparison that may not accurately reflect the complexities of AI infrastructure.
• The comparison to WW2 military spending is a dramatic claim that requires more substantial evidence.

Key Sources

• Martin Alderson — Author
• Morgan Stanley — Research Firm
• TSMC — Chip Manufacturer
• NVIDIA — AI Chip Company
• Broadcom — Chip Manufacturer
• AMD — AI Chip Company
• Sam Altman — CEO of OpenAI