So it started with apps, and now it's into AI?
That's right. We took that end-to-end principle and applied it to AI. If you're trying to deliver value with AI, you can buy a bunch of different components from different vendors and try to put them together yourself, or you can look for an integrated system where everything you need is already connected.
What are the different components one can put together to make a full AI stack?
An intentional AI stack requires an integrated combination of layers to get the job done: include the infrastructure, the AI model, the orchestration platform and user environments. At Google, we've deliberately invested in every single layer. We provide computing platforms such as Tensor Processing Units (TPUs), edge models developed by Google DeepMind such as the Gemini family of models, the Gemini Enterprise Agent Platform and areas people use every day, such as Maps and Gmail. We've done the hunting for you and packed all the necessary items inside the box.
Did we know we wanted to have a full-stack approach back when Google started working on AI?
It was a deliberate strategy, spanning decades. For example, our bet on custom TPUs is over 10 years old. We realized early on that there is a huge advantage to having our own supply chain and raw infrastructure when working on the world's most important Internet services. Managing that chain across the stack allows us to deliver a level of service, performance and reliability that is very difficult to achieve when you're on multiple teams.
On the other hand, is it accepting full-stack platform builders somehow?
That is a very valid concern, but locking people up is not consistent with our ethics. No company does open source like Google; we always provide the underlying technology and source code that the entire industry depends on.
We like to describe our AI platform as “visionary but scalable” and “batteries included” — meaning everything you need to build and run an app is ready out of the box. However, if you want to use another company's AI model instead of Gemini, or connect a different software instead of Google Workspace, you can connect that directly. We want you to use our products every day based on the integrity of our platform, not because we forced you to make a closed choice.
