Surface Laptop Ultra Pre-Order: Price, Release Date & Specs
Pre-orders for the Surface Laptop Ultra opened yesterday, alongside Microsoft's far pricier Surface RTX Spark Dev Box, as part of a broader Windows and Surface announcement (Microsoft, yesterday). Both machines ship October 16, the same day as companion RTX Spark PCs from Asus, Dell, HP, Lenovo, and MSI, Tom's Hardware reported yesterday.
Pricing starts at $2,599 for the Surface Laptop Ultra and $5,999 for the Surface RTX Spark Dev Box, per the same Tom's Hardware report. Both figures are listed in U.S. dollars, but that alone doesn't confirm a U.S.-only launch. The cited coverage doesn't break down regional storefronts, so shoppers outside the U.S. should check their local Microsoft Store page before assuming the same price or date applies.
Microsoft is launching the Surface Laptop Ultra as part of Nvidia's new RTX Spark PC platform. An earlier Microsoft description of that platform lists ceilings of up to 6,144 Blackwell RTX cores, 20 Arm CPU cores, 128GB of unified memory, and one petaflop of AI-optimized performance (Microsoft's EMEA announcement, earlier this year). Microsoft has not, in the launch reporting cited here, mapped those maximums to either the Laptop Ultra or the Dev Box.
The launch establishes the starting prices and the October 16 shipping date, but the cited coverage doesn't identify the memory, storage, display, or battery configuration attached to either price.
Surface Laptop Ultra pre-order: price and release date
Buyers can reserve a Surface Laptop Ultra for $2,599 right now, with delivery set for October 16. The Surface RTX Spark Dev Box pre-order carries a $5,999 starting price and the same ship date, alongside the batch of third-party RTX Spark desktops and laptops from Asus, Dell, HP, Lenovo, and MSI, Tom's Hardware reported.
Tom's Hardware's launch report did not identify a display size, memory capacity, or storage tier tied to either starting price. That's a gap worth checking directly on the product page rather than assuming.
Surface Laptop Ultra vs. Surface RTX Spark Dev Box: who each is built for
Microsoft frames the Laptop Ultra as a portable machine "designed for those creating the world to come" and aimed at creative professionals, according to its EMEA announcement earlier this year. A presenter at the launch event said the 4.5-pound chassis has the brightest peak HDR on a laptop, a company claim rather than an independently measured result, and described a magnetic USB-C charging system Microsoft calls Magnetic Connect, per Tom's Hardware.
The Dev Box costs more than double that price, and the reporting reviewed here doesn't include a chassis, port layout, or physical design to match.
The Dev Box is separate from Nvidia's DGX Spark and DGX Station for Windows. Microsoft's Davuluri described that system as bringing "last quarter's frontier to your desktop," built around Nvidia's GB300 Grace Blackwell Ultra Desktop Superchip and aimed at AI models with up to a trillion parameters, per Tom's Hardware and Microsoft's EMEA announcement. Microsoft said it aims to reach that tier this year, a target well past the hardware shipping October 16.
What Microsoft's AI benchmarks do not establish
Microsoft compared RTX Spark to Apple's M5 Pro chip, citing up to 2.1 times faster time-to-first-token, 4.3 times faster image generation, and 6.2 times faster video generation, Tom's Hardware reported. Those are Microsoft's own comparison numbers from the launch event, not figures produced by independent testing.
In Microsoft's demonstration at the same event, DeepSeek V4 Flash ran locally using 1.6-bit quantization and 60GB of memory, according to Tom's Hardware. That shows what the RTX Spark platform handled under Microsoft's controlled demo conditions, not what memory tier ships at the $2,599 or $5,999 starting price. Buyers with a specific local-model memory requirement should verify that detail before ordering, since the cited report doesn't identify the base configuration's usable memory.
For developers running existing x86 software, Microsoft says its Prism emulator is optimized for RTX Spark PCs, per its EMEA announcement. That's a company claim about optimization work, not a published benchmark comparing emulated apps against native Arm versions.
Software features tied to the launch
Windows ML is becoming the runtime Microsoft uses to deploy AI models across a PC's GPU, NPU, and CPU, and Llama.cpp is arriving inside Windows ML at launch, Tom's Hardware reported. Separately, Nvidia is bringing its OpenShell tool to Windows, leaning on the platform's newer security protections, per Microsoft's EMEA announcement.
Microsoft's Davuluri detailed Microsoft Execution Containers, OS-level sandboxes that control what data AI agents can access, which are moving to general availability, per Tom's Hardware. Davuluri also described a push toward "hybrid intelligence," splitting AI tasks between local and cloud agents to cut token usage, and said Windows will require agents to label which actions they've taken.
Microsoft's HydraFusion, which balances local and cloud AI agents, is scheduled to move from GitHub to Windows on October 15, a day before the hardware ships, covering GitHub Copilot, the CLI, and VS Code, and using new models including MAI Code 1.1 Flash and Nvidia's Nemotron, per Tom's Hardware.
Microsoft's Jacob Andreou demonstrated an updated Copilot that can read local files, run local models, and take approved actions, including drafting and sending an email after the user signed off, per Tom's Hardware. Andreou said those features are arriving over the next few months, so they should not be treated as confirmed launch-day features.
Gaming support follows the same pattern. A presenter at the launch event said hundreds of AAA and indie titles will be playable day one, with Call of Duty arriving on RTX Spark hardware in 2027, according to Tom's Hardware. No title list or frame-rate figures accompanied that claim, so it remains a Microsoft/Nvidia presentation claim until independent testing establishes compatibility and performance.
What to check before you order
Check the official product page and checkout configuration before ordering either machine, since the cited coverage doesn't tie memory, storage, or display specs to either starting price. Buyers planning around a specific local AI model should confirm usable unified memory on their exact configuration rather than extrapolate from Microsoft's demo.
Developers should check whether their existing tools run natively on Arm or depend on Prism emulation, since no comparison between the two has been published. Buyers considering either machine have reason to wait for the exact configuration pages and independent reviews of battery life, thermals, and real-world AI performance once units start shipping October 16.
Comments
Be the first, drop a comment!