{"slug": "nvidia-and-microsoft-put-a-data-center-chip-inside-a-2599-laptop", "title": "Nvidia and Microsoft put a data center chip inside a $2,599 laptop", "summary": "Microsoft and Nvidia launched the Surface Laptop Ultra on October 7 in San Francisco, a 15-inch notebook built around Nvidia's RTX Spark superchip starting at $2,599 and shipping October 16, with roughly one petaflop of FP4 AI performance and up to 128GB of unified memory. Nvidia CEO Jensen Huang said on stage, \"The PC is being reinvented,\" claiming the machine can run 120-billion-parameter models locally without cloud inference. Lenovo, Asus, Dell, MSI and HP are shipping their own RTX Spark laptops — the Yoga 9n, ProArt P14, XPS 16, Prestige N16 and OmniBook Ultra 16 — all beginning to ship October 16, while Microsoft opened preorders for the $5,999 Surface RTX Spark Dev Box.", "body_md": "*Nvidia just put a petaflop of AI compute into a laptop you can buy next week, and it runs 120 billion parameter models without touching the cloud. That's the kind of local horsepower cloud-inference startups were betting nobody would have on their desk.*\n\nOn October 7, in San Francisco, Microsoft CEO Satya Nadella, Windows and Surface chief Pavan Davuluri, and Nvidia CEO Jensen Huang stood on the same stage to launch the Surface Laptop Ultra, a 15-inch notebook built around Nvidia's new RTX Spark superchip. The entry model starts at $2,599. It ships October 16.\n\nThat's not a prototype or a developer kit with a waitlist. It's a shipping date, nine days out, with a real price tag attached. According to Engadget's live coverage of the event, the Surface Laptop Ultra comes in two RTX Spark configurations: an 18-core version with 5,120 GPU cores, and a 20-core version with 6,144 GPU cores for heavier workloads. Both pair Nvidia's Grace Blackwell design, a 20-core Arm CPU fused to a Blackwell GPU on one module, with a unified memory pool that scales up to 128GB and roughly one petaflop of FP4 AI performance, as Notebookcheck and multiple outlets covering the launch reported.\n\nHuang's framing on stage was blunt. \"The PC is being reinvented,\" he said, according to coverage of the keynote. \"For forty years, you launched apps. Click. Type. With RTX Spark and Microsoft Windows, you ask, and the PC does the work.\" He went further: \"RTX Spark brings everything Nvidia has built, CUDA, RTX, our AI platform, into a single superchip. Local agents. Frontier models. Creative workflows. RTX games. All on a laptop.\"\n\nMost laptop AI chips today are built to run small assistant models, the kind that summarize an email or caption a photo. A 120-billion-parameter model is a different class entirely, closer to what companies currently rent GPU time in the cloud to run. If that workload now fits inside a $2,599 machine with no API calls and no per-token bill, the economics change for anyone who built a product on charging customers for inference they used to have to buy from a data center.\n\n[Nvidia brings local AI computing into the Windows PC fight](https://startupfortune.com/nvidia-brings-local-ai-computing-into-the-windows-pc-fight/)\n\nNvidia’s RTX Spark brings high-end local AI compute into Windows PCs, backed by Microsoft’s new Surface RTX Spark Dev Box. The real test is whether developers get useful local inference workflows, not just another premium hardware cycle. - [nvidia RTX Spark local AI computing for Windows](https://startupfortune.com/nvidia-brings-local-ai-computing-into-the-windows-pc-fight/) - [how to build AI agents on Windows PCs](https://startupfortune.com/nvidia-brings-local-ai-computing-into-the-windows-pc-fight/)\n\nThat's the real threat buried in the spec sheet. A huge share of AI startups today are thin wrappers around a cloud model: they take a cut of the inference cost and charge the customer a markup. If the customer's own laptop can run a 120-billion-parameter model locally, for free, after the hardware is paid for once, the wrapper's reason to exist gets thinner. Nvidia isn't trying to kill that business on purpose. It's just that extending its CUDA moat down into consumer hardware happens to gut it as a side effect.\n\nNvidia isn't alone in building for this moment. Lenovo, Asus, Dell, MSI and HP are all shipping their own RTX Spark laptops, the Yoga 9n, the ProArt P14, the XPS 16, the Prestige N16, and the OmniBook Ultra 16, all open for preorder alongside the Surface Laptop Ultra and all beginning to ship October 16. That's not one company testing a niche product. It's five PC makers plus Microsoft, all betting the same chip platform is worth building a whole laptop line around at once.\n\nFor developers who want more raw compute than a laptop chassis allows, Microsoft also opened preorders for the Surface RTX Spark Dev Box, a desk-bound version of the same silicon, starting at $5,999. That price gap, $2,599 for something you carry versus $5,999 for something that sits on a desk, tells you plainly which product Nvidia and Microsoft think is the mass-market bet and which is the workstation for people already building on top of it.\n\nNone of this is final-form yet. RTX Spark was only unveiled in prototype form back on May 31, at Computex in Taipei, and Microsoft spent the summer declining to confirm a price or ship date. October 7 is the day that changed: real numbers, a real shelf date, real competing laptops from five other manufacturers. Whether 128GB of unified memory and a petaflop of local compute actually displaces cloud inference at scale is still an open question nobody can answer from a keynote stage. But the chip is real, the price is real, and in nine days, so is the laptop.\n\n**Also read:** [Perplexity open sources its ColBERT style search models under MIT license](https://startupfortune.com/perplexity-open-sources-its-colbert-style-search-models-under-mit-license/) • [Vitalik Buterin tells crypto builders AI is a privacy and security nightmare](https://startupfortune.com/vitalik-buterin-tells-crypto-builders-ai-is-a-privacy-and-security-nightmare/) • [Google, Meta and the US government put $1.8 billion behind a virtual human cell](https://startupfortune.com/google-meta-and-the-us-government-put-18-billion-behind-a-virtual-human-cell/)\n\n*This article is posted in [AI News](https://startupfortune.com/category/ai/), check it out for more related stories.*\n\n[Nvidia brings RTX Spark to Windows laptops.](https://startupfortune.com/nvidia-brings-rtx-spark-to-windows-laptops/)\n\nNvidia's RTX Spark brings an Arm-based AI superchip to Windows laptops, with Microsoft Surface Laptop Ultra leading the first wave. The move gives Nvidia a more direct role in the PC processor market and pressures Intel and AMD at the high-performance edge of Windows computing. - [Nvidia RTX Spark Windows laptop processor](https://startupfortune.com/nvidia-brings-rtx-spark-to-windows-laptops/) - [Arm based processor challenging Intel AMD dominance](https://startupfortune.com/nvidia-brings-rtx-spark-to-windows-laptops/)\n\n## Join the discussion\n\n[Open in the community →](https://startupfortune.com/community/)\n\nAlmost there. Sign in and your reply posts straight away.", "url": "https://wpnews.pro/news/nvidia-and-microsoft-put-a-data-center-chip-inside-a-2599-laptop", "canonical_source": "https://startupfortune.com/nvidia-and-microsoft-put-a-data-center-chip-inside-a-2599-laptop/", "published_at": "2026-10-07 19:03:48+00:00", "updated_at": "2026-10-07 19:17:42.751565+00:00", "lang": "en", "topics": ["ai-chips", "ai-infrastructure", "ai-products", "large-language-models"], "entities": ["Nvidia", "Microsoft", "Surface Laptop Ultra", "RTX Spark", "Jensen Huang", "Satya Nadella", "Pavan Davuluri", "Surface RTX Spark Dev Box"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/nvidia-and-microsoft-put-a-data-center-chip-inside-a-2599-laptop", "markdown": "https://wpnews.pro/news/nvidia-and-microsoft-put-a-data-center-chip-inside-a-2599-laptop.md", "text": "https://wpnews.pro/news/nvidia-and-microsoft-put-a-data-center-chip-inside-a-2599-laptop.txt", "jsonld": "https://wpnews.pro/news/nvidia-and-microsoft-put-a-data-center-chip-inside-a-2599-laptop.jsonld"}}