TickerTain
$SPCX····$NVDA····$MSFT····$GOOGL····$ANTHROPIC·$TSLA····$CRWV····$NBIS····$HIMS····$TSM····$ASML····$FIGURE·$LYFT····$UBER····$SNOW····$META····$ELEVENLABS·$CURSOR·$COGNITION·$OPENAI·$AAPL····$RKLB····$XAI·$AMZN····$AMD····$000660.KS····$INTC····
← newsroom
Semiconductors · Three pillars of 3x compute scaling all hitting hard ceilings by 2025-2030
now playing · Semiconductors
AI Infrastructuretailwindscore 9/10dwarkesh patel
Compute prices could 10-15x as AI revenue 10x's while hardware only 3x's
Frontier lab revenue is compounding at 10x/year while compute supply only scales 3x/year (Moore's Law 1.4x + new fabs 1.2x + wafer reallocation 1.8x). This structural imbalance forces eithe…
Semiconductorstailwindscore 9/10dwarkesh patel
Semiconductor supply inelasticity creates structural bottleneck for AI compute scaling
The 3x annual compute growth relies on three components (Moore's Law 1.4x, new fabs 1.2x, wafer reallocation 1.8x) all hitting physical limits by 2025-2030, making supply unable to absorb d…
Semiconductorsheadwindscore 9/10dwarkesh patel
Three pillars of 3x compute scaling all hitting hard ceilings by 2025-2030
The 3x annual compute growth decomposes to: 1.4x from Moore's Law (slowing, miracle to sustain), 1.2x from new fabs (bottlenecked by ASML EUV output through 2030+), 1.8x from wafer realloca…
AI Infrastructuretailwindscore 8/10dwarkesh patel
Compute prices may 10-15x as AI models approach human-level capabilities
As AI models become more capable, they can monetize compute far more effectively, driving up compute prices despite 3x annual supply growth. The marginal value of compute could reach $250K/…
AI Economics & Business Modelstailwindscore 8/10dwarkesh patel
Alchian-Allen effect gives efficient model makers pricing power on expensive compute
As compute costs rise, labs with more token-efficient models can charge large premiums because weaker models burn more expensive compute for same results, creating winner-take-most dynamics…
AI Economics & Business Modelstailwindscore 8/10dwarkesh patel
Alchian-Allen effect: expensive compute favors most efficient models, enabling premium pricing
When compute becomes scarce and expensive ($20/hr for H100), labs with more token-efficient models gain compounding advantage — they burn fewer tokens per task, effectively 'creating comput…
Neoclouds & Cloud Computingtailwindscore 7/10dwarkesh patel
SpaceX and specialized neoclouds capturing 2x spot premiums for secured, efficient GPU clusters
Frontier labs (Google, Anthropic) are paying 2x spot prices ($900M/month for 110K GPUs) to neocloud providers like SpaceX for guaranteed capacity, efficiency, and security — signaling a str…
AI Bubble / Capex Debateriskscore 7/10dwarkesh patel
Compute scarcity differs from commodity markets due to inelastic supply and lack of substitutes
Unlike the Simon-Ehrlich bet on metals where innovation increased supply, compute supply is far less elastic and substitutable, suggesting current scarcity may persist rather than being sol…