The Ai Universe
- Shakeel Awan
- Jun 26
- 1 min read
The 5 layers
The current 5 layers of Ai are energy, chips, infrastructure, language models and applications.
This is how the stack is made up, to go from the physical power needed to run AI, all the way up to the products people actually use.
1 Energy
The energy is provided through the use of electricity to power data centres and AI systems.
Energy must grow from its current levels to fulfil exceptional demand.
2 Microchips
The second level is microchips.
Specialized hardware like GPUs that do the heavy computation.
This is the brain of Ai.
3 Infrastructure
The datacentres, systems and software.
Data centers combine networking, electricity, power cooling systems, storage and software.
The systems, the software, train, run and serves AI models, as well as allow multiple chips to work together.
4 Ai Models
Ai Models are the trained AI systems, such as large language models.
These large language models are developed by Ai corporations.
Large language models generate predictions or answers.
5 Applications
Applications are the public view of Ai systems;
these are the digital systems that sit on top of Ai and provide access to Ai systems.
These systems include co-pilots, chatbots and automation products.
Conclusion
The dissection of Ai gives us an indication of the commitment required to build Ai infrastructure.
The 5 layers form a pyramid; at the top, we have public and commercial adoption of Ai.
Then beneath that we have the support structures that fulfil that demand.
By Shakeel Awan, at Xecology.
Comments