Surface Laptop Ultra Opens Preorders With RTX Spark and Up to 128GB Memory

Microsoft has opened preorders for the Surface Laptop Ultra, a new Windows laptop built around NVIDIA's RTX Spark platform rather than a conventional Intel or AMD processor paired with a separate GeForce graphics card. The most striking specification is up to 128 GB of unified memory, which Microsoft says gives developers room to run some AI models with more than 120 billion parameters locally.

The laptop was unveiled on October 7, 2026, alongside a compact Surface RTX Spark Dev Box. Microsoft lists US pricing for the Laptop Ultra from approximately ₱163,332.67 (US$2,599.99) on its US product page. This peso figure is an approximate conversion at ₱62.8205 per dollar on October 9, not an official Philippine suggested retail price. Availability is scheduled to begin October 16 in supported markets. A confirmed Philippine price and local preorder date were not available in the company's announcement.

This is an interesting moment for PC buyers because the pitch is changing. Instead of focusing only on faster gaming or office apps, Microsoft wants one machine that can edit video, write code, run local AI tools and still handle games after work. Whether it is worth the premium price depends heavily on which of those jobs you actually do.

RTX Spark combines NVIDIA graphics with a Grace CPU

The Laptop Ultra uses NVIDIA's RTX Spark Superchip, combining a Blackwell RTX GPU with up to 6,144 GPU cores and a Grace CPU with up to 20 cores. The architecture uses a shared pool of system memory rather than dividing the machine into conventional CPU RAM and a separate dedicated GPU VRAM allocation.

That memory design is a major part of the story. A high-memory configuration can make it possible to load AI models that would not fit into the dedicated memory of many gaming laptops. However, the advertised 128 GB is the maximum system-memory configuration, not a promise that every model has 128 GB or that the GPU can use all 128 GB without overhead. Microsoft explicitly notes that some memory is reserved for the system and the amount addressable by the GPU varies by workload.

NVIDIA advertises up to one petaflop of FP4 AI compute on RTX Spark. That is a theoretical low-precision figure that uses sparsity and should not be read as a universal speed rating. Real-world model performance will depend on quantization, model architecture, context length, memory use and the software stack. It also cannot be directly compared with a GPU's gaming frame rates.

Microsoft's Surface announcement positions the machine for creative professionals, developers and AI builders. NVIDIA's October 7 update emphasizes CUDA and the ability to move AI projects between laptops, desktops and larger NVIDIA systems.

What the laptop actually includes

Microsoft specifies a 15-inch PixelSense Ultra touchscreen with a 120 Hz refresh rate, a chassis under 18 mm thick and weighing under 4.5 pounds. The display is designed for creators, with HDR and wide-color capabilities, while NVIDIA's graphics platform provides ray tracing, DLSS and Reflex in supported games.

There are practical details beyond the headline chip. Microsoft has introduced a magnetic USB-C charging connection that can detach when the cable is pulled, while still functioning as a normal USB-C port for data and accessories. The laptop also includes two additional USB-C ports, HDMI, USB-A, an SD card reader and a headphone jack. That is a more useful selection for creators than a design that requires a dongle for every camera card or external monitor.

Microsoft says the storage is user-removable, with capacities up to 2 TB. Replacement parts and repair options vary by market. The company also says it redesigned the cooling system for sustained performance, but its claimed thermal improvements should not be confused with independent benchmarks. The laptop's full battery life, fan noise and performance under extended AI workloads still need outside testing.

Can it really run large AI models offline?

The hardware is clearly intended for local inference and AI development. Microsoft says models exceeding 120 billion parameters can run locally on suitable configurations, and NVIDIA points to its CUDA software stack as a key advantage. Local operation can reduce reliance on remote inference services for some tasks and may help keep selected data on the user's device.

That does not mean every model of that size will run quickly, fit at full precision or behave identically to a cloud service. Model format, memory footprint, runtime support and context length matter. Nor does local processing automatically make a workflow private: connected apps, plugins, synchronization and cloud-backed assistants may still send information online. Buyers should check the actual software path they plan to use.

For programmers, Microsoft highlights local GitHub Copilot workflows, testing and code review. The appeal is the ability to keep development tools and a larger model running side by side without depending on a remote GPU for every experiment. Whether that benefit justifies the cost depends on how often the user actually needs local large-model inference rather than a cloud API or a smaller model on an existing PC.

A separate Dev Box targets more demanding developers

Microsoft also announced the Surface RTX Spark Dev Box, a compact Windows 11 Pro machine aimed at developers and AI engineers. Its US product page lists a starting price of approximately ₱376,922.37 (US$5,999.99), converted at the same October 9 reference rate. Microsoft says it is available for US-only preorders, with shipping beginning in November 2026. Neither the dollar price nor the converted peso estimate is a Philippine MSRP.

The Dev Box offers up to 128 GB unified memory and up to one petaflop of advertised FP4 AI performance. It comes with a development-focused Windows setup, including Visual Studio Code, GitHub tooling, WSL and PowerShell 7. That makes it a different proposition from the Laptop Ultra: it is designed to sit on a desk and handle repeated local workloads rather than travel with its owner.

See the official Dev Box specifications.

What about gaming and existing Windows apps?

The Surface Laptop Ultra can play supported PC games using RTX features, but its Grace processor also means buyers should pay attention to Windows on Arm compatibility. A title that works on a traditional x86 Windows gaming laptop should not automatically be assumed to run with the same performance or anti-cheat compatibility on this system.

Microsoft describes a growing Windows on Arm game catalog and lists RTX gaming features, but does not provide an independent game-by-game compatibility report in its announcement. Gamers should verify the specific titles, launchers, peripherals and creative applications they depend on before spending this much. The same caution applies to niche development tools that require particular drivers or native x86 components.

There is no reason to infer from the one-petaflop AI figure that this machine will outperform a high-end discrete gaming GPU in every game. AI inference and game rendering stress different parts of the system, and comparisons need matched benchmarks, power settings and tested games.

Philippine price and availability: what is confirmed

Microsoft's US product listing shows a starting US price of ₱163,332.67 (US$2,599.99), expressed here as an approximate peso conversion. The October 9, 2026 exchange rate used for this estimate is US$1 = ₱62.8205. Import taxes, card fees, shipping, retailer margins and warranty terms could make a Philippine retail price substantially different. Microsoft has not supplied a verified local Philippine MSRP or launch schedule in the cited announcement.

The Laptop Ultra is due to become available October 16 in supported markets. The Dev Box is a separate US-only preorder product scheduled to ship in November. Neither statement confirms a Philippine store listing, so readers should avoid treating US preorder buttons as evidence of local availability.

The practical takeaway is that RTX Spark is now moving from an AI PC concept to hardware with preorder dates and retail prices. For developers who genuinely need large local models and CUDA tooling, Surface Laptop Ultra is worth watching. For most students and gamers, the sensible move is to wait for independent benchmarks, application compatibility reports and an actual Philippine price before deciding whether the new platform offers enough value over existing laptops or desktops.

Sources and verification

This report is based on Microsoft's October 7 device announcement, its Laptop Ultra product page and Dev Box product page, plus NVIDIA's RTX Spark announcement. Currency figures are approximate conversions using an October 9 USD/PHP reference rate, not Philippine retail quotations.