‘Surface Laptop Ultra’ pre-orders begin... Supports up to 128GB memory
Process AI tasks by dividing PC and cloud… Strengthen security and reduce usage costs
Microsoft (MS) has started targeting the market by starting pre-orders for high-performance AI PCs equipped with NVIDIA chips. The idea is to reduce usage costs by moving some of the artificial intelligence (AI) work that was processed in the cloud to PCs and expand the scope of users' ability to directly manage their own data.
Microsoft began pre-ordering the ‘Surface Laptop Ultra’ on the 7th (local time). This product is equipped with the ‘RTX Spark’ chip jointly developed by NVIDIA and MediaTek and supports up to 128GB of integrated memory. MS explained that some large AI models can be run on the device and used for code writing, review, and model testing. The release date is the 16th.
The computing power inside these devices is the basis of the ‘hybrid intelligence’ strategy presented by Microsoft. Tasks that can be processed on a PC are performed on the device, and tasks that require more calculations are left to the cloud. The purpose is to reduce the burden of cloud use caused by repeatedly calling AI, but to utilize the cloud's processing power when necessary.
This strategy also applies to AI assistant Co-Pilot. The plan is to obtain the user's permission to understand the files and work context stored on the PC, and to utilize the AI model inside the device when necessary while organizing files or resolving device problems. Related features will be provided sequentially to CoPilot+ PC over the coming months.
As the scope for AI to read files and directly perform tasks expands, a device to control access rights has also been established. MS announced on this day that it will officially provide ‘Microsoft Executable Container (MXC)’ in Windows 11. This is a technology that applies the rules in the actual execution process when a company determines the scope of file and network access allowed to the AI agent. The purpose is to restrict access to unauthorized data or systems while entrusting tasks to AI.
The product line that supports this also expands. Asus, Dell, HP, Lenovo, and MSI have also begun taking pre-orders for RTX Spark-based PCs. Microsoft has also begun accepting pre-orders for the Surface RTX Spark Dev Box, a compact desktop for developers and AI engineers, and is scheduled to ship in the US in November. The idea is to enable AI models to be run and tested in a variety of work environments through laptops and desktops.
High equipment prices are a burden for expansion. The basic version of the Surface Laptop Ultra with 24GB of memory is sold in the US starting at $2,599.99. Reuters reported an expert assessment that the increased cost of purchasing equipment due to rising memory prices could be an obstacle to the spread of running AI directly on PCs. In addition to the expectation that cloud usage fees can be reduced, the initial equipment purchase cost must also be considered.
