Nvidia CEO Jensen Huang explains that AI development relies on a five-layer structure—energy, chips, cloud infrastructure, AI models, and applications—each requiring massive investment, with trillions more needed to fully build out the ecosystem. He emphasizes that while public focus is on AI models, the real economic value and rapid innovation are happening at the application layer, fueled by significant growth and funding across all foundational layers.
Nvidia CEO Jensen Huang explains that the development and deployment of artificial intelligence (AI) is best understood as a five-layer structure, or “cake.” At the foundation is the energy layer, which is essential because AI processes information and generates intelligence in real time, requiring substantial and reliable energy resources. This foundational need for energy underpins all other layers in the AI ecosystem.
The second layer consists of chips and computing infrastructure, which is Nvidia’s primary domain. This includes the manufacturing and deployment of advanced semiconductors and the hardware necessary to support AI computations. Above this is the cloud infrastructure layer, which provides the scalable computing resources and services that enable AI models to be trained and deployed at scale.
The fourth layer is the AI models themselves, which is where most public attention is focused. However, Huang emphasizes that these models cannot function without the robust infrastructure provided by the layers beneath them. The most critical and rapidly evolving layer, according to Huang, is the application layer at the very top. This is where AI delivers tangible economic benefits across industries such as financial services, healthcare, and manufacturing.
Huang highlights that the need for all these layers has triggered the largest infrastructure buildout in human history. While hundreds of billions of dollars have already been invested, he estimates that trillions more will be required to fully realize the potential of AI. This massive investment is justified by the need to process vast amounts of data and generate intelligence for a wide range of applications.
He points to significant growth and investment across all layers: energy, chip manufacturing (with companies like TSMC, Foxconn, and Micron expanding production), and the application layer, which is attracting record levels of venture capital. In 2024, much of this funding has gone to “API-native” companies in sectors like healthcare, robotics, manufacturing, and financial services, reflecting the fact that AI models have now reached a level of maturity that enables meaningful innovation and value creation on top of them.