Windows on ARM: Asus with Snapdragon X2 Extreme Edition, HP with NPU Turbo
First notebooks with Snapdragon X2 are coming from Asus and HP – and are immediately exceptions to the mass that will follow later.
Asus ZenBook A16 with Qualcomm Snapdragon X2 Elite Extreme.
(Image: Florian MĂĽssig / heise medien)
Among the first notebooks with Qualcomm's new processor generation Snapdragon X2 are the ZenBooks A14 and A16 from Asus and the HP models EliteBook X G2q and OmniBook Ultra G2q. And both have special features right from the start that other devices will lack.
The ZenBook A14, like its predecessor launched a year ago, is a particularly light 14-inch model that is now making the leap to the new CPU generation. The new ZenBook A16 not only adds a variant with a larger 16-inch screen to the series, but is also the first and so far only notebook to feature the top model Snapdragon X2 Elite Extreme.
Technically, this processor runs outside the rest of the X2 portfolio because it requires its own CPU version and thus customized mainboards. Its memory interface includes three instead of the usual two memory channels (192 instead of 128 bits), and moreover, the RAM is part of the CPU carrier. For this version, there is currently only exactly one CPU model, namely the 18-core X2E-96-100 with 48 GByte of integrated LPDDR5X RAM – meaning the Asus ZenBook A16 always has this on board.
Asus has not yet provided any information on price and sales launch.
NPU with 85 instead of 80 TOPS
HP, in turn, is not limiting its 14-inch EliteBook X G2q and OmniBook Ultra G2q models to the CPU selection previously revealed by Qualcomm, which is available to all manufacturers. Instead, the HP-exclusive models X2E-90-100 (18 CPU cores) and X2E-84-100 (12 CPU cores) will be used in the notebooks expected to be available from March. They are similar to the freely available variants X2E-88-100 and X2E-80-100, respectively, but have an enhanced AI unit (Neural Processing Unit, NPU) that adds to the already outstanding 80 TOPS (trillions of operations per second) of the regular X2 models: here there are 85 TOPS.
(Image:Â Florian MĂĽssig / heise medien)
When asked what users can expect from all this NPU computing power significantly above the 40 TOPS required by Copilot+, Microsoft's James Howell stated that in the future, multiple AI models will simply be able to run in parallel. For example, if someone wants to send a requested file via email during a Teams meeting (with an AI-softened background) but forgets to actually attach the file in the rush, they will no longer have to send a second email with an apology. Instead, a locally running language model will analyze the content of the email when you click send, and Outlook will prompt the user again if the AI detects that an attachment is mentioned but is missing.
White-Label Snapdragon
Given that Microsoft and Qualcomm share the same vision regarding NPUs and AI on notebooks and are demonstrating close cooperation, it is not to be expected that Qualcomm will invest much effort in supporting operating systems other than Windows. This realization led Tuxedo to scrap its project to launch a Linux notebook with Snapdragon X at the end of 2025.
(Image:Â Florian MĂĽssig / heise medien)
However, there is no shortage of white-label hardware that small local notebook providers can use as a basis for their device offerings: Qualcomm showed at CES which contract manufacturers (ODMs) are behind the reference systems shown so far and used for benchmarks. The notebooks come from Compal (KQX80, KQX81), Wistron (Oryon2 Clamshell), and Quanta (QM8), and the all-in-one PCs from Longcheer. It is possible that one or the other device may be encountered in the future under a different name in retail.
heise medien is an official media partner of CES 2026.
Empfohlener redaktioneller Inhalt
Mit Ihrer Zustimmung wird hier ein externer Preisvergleich (heise Preisvergleich) geladen.
Ich bin damit einverstanden, dass mir externe Inhalte angezeigt werden. Damit können personenbezogene Daten an Drittplattformen (heise Preisvergleich) übermittelt werden. Mehr dazu in unserer Datenschutzerklärung.
(mue)