AI Bubble Concerns Shift to Compute Infrastructure, Not Adoption
TL;DR. Market observers question if the current AI bubble is in compute infrastructure rather than in AI adoption itself. - The AI market opportunity is vast, fueled by productivity gains and labor substitution. - Speculation centers on whether compute will remain the dominant bottleneck. - If AI models become efficient and cheap faster than expected, infrastructure valuations could be at risk.
- The AI market is broadly right about AI's potential to increase productivity and substitute labor.
- The most aggressive market bet is on compute remaining the dominant bottleneck, benefiting GPU, memory, and cloud infrastructure suppliers.
- A scenario where AI models become good, cheap, and open faster than anticipated could devalue current infrastructure investments.
- The 'bubble' may be in the assumption that current compute scarcity will persist long-term.
Sources
- If there is an AI bubble, where is it? — gregoryap.substack.com