The Sequence

AI’s Sixth Layer Is Finance

This opinion piece argues that Jensen Huang’s five-layer model of the AI economy (energy, chips, infrastructure, models, applications) is incomplete because it omits finance. The author proposes a sixth layer—finance or the balance sheet—underpinning the stack, noting that capital allocation, depreciation, debt covenants, and operating costs (power, cooling, data-center infrastructure) are essential to understanding AI’s economics. The article frames AI infrastructure as an industrial system where capital flows up the stack while revenue flows back down, shaping who can build and operate large-scale AI systems.

AI Energy Scaling Laws and Infrastructure

An opinion essay arguing that AI scaling is constrained not only by model parameters but by physical energy infrastructure. The piece emphasizes that AI workloads depend on substations, cooling systems, transmission networks, and power plants — and that most electrical energy consumed by clusters is ultimately emitted as heat. The author frames the next major scaling law as society’s capacity to convert primary energy (photons, motion, nuclear) into useful intelligence, highlighting operational and physical realities behind seemingly weightless cloud AI experiences.

Four New Frontier AI Model Releases

The article summarizes four recent AI model announcements: DeepSeek shipped the general-availability V4‑Pro, Z.ai introduced GLM‑5.3, and NVIDIA released Nemotron 3.5 Lightning alongside NeMo Switchyard. The newsletter notes benchmark tables accompanying the releases and provides a concise technical discussion of the models to keep readers current on frontier LLM developments. The piece is a short, analytical roundup of these model releases aimed at readers tracking advances in large language models and inference infrastructure.