The demand for AI is rapidly outstripping the capacity of physical infrastructure. Data center growth is colliding with limitations in power grids, water access, and permitting, making these real-world resources the ultimate gatekeepers for the expansion of AI capabilities.

Progress in AI isn't a smooth, continuous line. Just as Moore's Law required discrete inventions, AI scaling relies on paradigm shifts. The current Transformer+RL approach may hit diminishing returns, and an AI trained within this paradigm is unlikely to discover the next fundamental breakthrough required to maintain progress.

The current 3x annual growth in AI compute is not easily accelerated and may be unsustainable. It is fundamentally constrained by the slowing of Moore's Law (1.4x), the fixed production rate of ASML's EUV machines for new fabs (1.2x), and the near-total absorption of leading-edge wafer capacity from other sectors (1.8x).

Dario Amodei simplifies the complex concept of AI scaling laws with an analogy: just as a chemical reaction needs ingredients in proportion to create fire, AI needs data, compute, and model size in proportion to create the product of intelligence.

The relationship between computing power and AI model capability is not linear. According to established 'scaling laws,' a tenfold increase in the compute used for training large language models (LLMs) results in roughly a doubling of the model's capabilities, highlighting the immense resources required for incremental progress.

The focus in AI has evolved from rapid software capability gains to the physical constraints of its adoption. The demand for compute power is expected to significantly outstrip supply, making infrastructure—not algorithms—the defining bottleneck for future growth.

With past shifts like the internet or mobile, we understood the physical constraints (e.g., modem speeds, battery life). With generative AI, we lack a theoretical understanding of its scaling potential, making it impossible to forecast its ultimate capabilities beyond "vibes-based" guesses from experts.

Every layer of the AI supply chain is constrained, from energy and data centers to turbines, transformers, and rare earth minerals. This is a shift from software limitations to hard physical constraints. As a result, the price of intelligence may stop decreasing and could even rise.

Contrary to the narrative that model performance is plateauing, Demis Hassabis states that while returns from scaling are no longer exponential, they remain 'very substantial.' Frontier labs continue to see significant gains from increasing model size and compute, suggesting the current AI paradigm is not yet exhausted.