What is an AI computer? Intel and Microsoft have finally given their definition

The industry has yet to reach a single definition of an AI computer. Numerous companies, including Intel, AMD, Apple, and soon Qualcomm with its X Elite chips, have developed silicon with special AI accelerators that are placed on the chip alongside standard CPU and GPU cores. However, each of them has its own idea of what an AI computer is.

Microsoft and Intel have presented their vision of the term: an AI computer will be equipped with a neural processing unit (NPU), a CPU and GPU that support Microsoft Copilot, and a physical Copilot key directly on the keyboard that replaces the second Windows key on the right side of the keyboard, Tom’s Hardware reports.

Copilot is an AI chatbot based on LLM that is currently being implemented in newer versions of Windows 11. For now, it runs on cloud services, but the company plans to enable local data processing to improve performance and responsiveness. This definition means that Meteor Lake and Ryzen laptops that come without the Copilot key do not actually meet Microsoft’s official criteria.

Intel plans to ship more than 100 million AI-enabled PCs by the end of 2025. The company is already working with more than 100 AI vendors for PC platforms and plans to bring more than 300 AI-enabled applications to market by the end of 2024.

The struggle for control of the AI computing market will intensify in the coming years – according to Canalys forecasts, 19% of all computers on the market will be equipped with artificial intelligence in 2024, and by 2027 this figure will increase to 60%.

AI models also have fairly high requirements for RAM and speed, with the former allowing for larger and more accurate models and the latter providing greater performance. AI models come in all shapes and sizes, and Intel says that memory will be a key issue when launching LLM, with some workloads requiring 16 GB and even 32 GB depending on the types of models used.