The moment many have been waiting years for has arrived. Nvidia has long made graphics cards that powered the Windows PC ecosystem for decades—now it wants to control the whole thing with “superchips,” starting with the RTX Spark.
Announced over the weekend at the Computex tech expo in Taiwan, RTX Spark chips combine unified memory, RTX graphics, and the new part: the N1 CPU. Nvidia already owns the entire world of AI processed in data centers with its GPUs. But now it’s making its play for locally run AI too. It’s just a tease of what’s to come right now, but these are the first Windows devices that may actually live up to the overused “AI PC” name.
The Fake AI PC
Microsoft has been talking up the idea of “AI PCs” since 2024, but it has never felt real. Sure, its Copilot+ PCs had built-in neural processing units (NPUs) and started with 16 GB of RAM, but they didn’t have the performance to run large language models locally any more than your phone. Rather than build the hype machine around AI, Microsoft’s early promises about the next era of the PC felt empty.
While they need to be tested, and prices are still undetermined, these new Nvidia laptops look truly like real AI PCs. The combination of unified memory up to 128 GB, an efficient Arm-based CPU, and the company’s trademark RTX graphics cards gives you a computer unheard of outside the MacBook Pro. This competition is important as the MacBook Pro (or Mac desktops) was the only serious option for AI enthusiasts looking to run foundation-level models locally.
Courtesy of Microsoft
Nvidia is opening up its technology to the usual collaborators in the Windows space, such as HP, Asus, Dell, Lenovo, and others. Microsoft’s own Surface Laptop Ultra, though, might be the pinnacle of this new breed of PC. It’s a proper MacBook Pro alternative, featuring a 15-inch Mini-LED display and several ports in a package that appears to be similar in size. Microsoft hasn’t had a performance-driven Surface device in years, and this one comes ahead of a rumored “MacBook Pro Ultra” coming later this year.
I’d be lying if I said I wasn’t excited. One of the issues with Windows laptops that try to compete with the MacBook Pro is battery life. Because they always need to rely on discrete graphics cards (usually from Nvidia), things like battery life and fan noise tend to nose-dive. Only this year, with the release of Intel’s new Core Ultra Series 3 chips, have we seen laptops like the Dell XPS 14 that can balance GPU performance and battery life closer to the level of the MacBook Pro. But those configurations max out at 64 GB of memory.
It’s not just about more memory. It’s also the fact that the graphics will be up to as powerful as a discrete RTX 5070. And it’s also the software layer (known as CUDA) that allows developers to access the system’s GPU cores. Nvidia has built a highly developed AI platform around CUDA because of its use in data centers. Bringing that AI processing prowess to local PCs could mean AI performance that pulls far more out of the hardware than current computers do. Mac development has grown rapidly over the past year, but it’s still far behind the scale that Nvidia could do with RTX Spark.
Sparking a Revolution
Don’t be fooled into thinking these will be priced affordably. Some reports estimate that prices for high-end configurations of RTX Spark laptops will be over $4,000. That shouldn’t come as a shock, as that’s what a similarly configured MacBook Pro costs these days.
It’s no longer difficult to envision a near future in which local AI models are routinely used for inference across a wide range of work projects. As agentic models become easier to use and more streamlined, using them locally will increasingly be preferred for privacy reasons too. We’re already seeing the demand skyrocket for the Mac Mini, which is facing long shipping times that Apple attributes to surprisingly fast AI adoption. RTX Spark has its eyes set on the Mac Mini as well, with several small-form-factor (SFF) compact desktop PCs on the way.
Read the full article here