Holy moly, did you see Sean Hollister's piece on The Verge about Nvidia laying out their whole next-gen roadmap? Jensen Huang dropped some genuinely jaw-dropping news at Computex 2026 in Taipei—he confirmed that they're planning N2X *and* N3X chips for RTX Spark right now! I love this because it means Nvidia isn't just testing the waters with a one-off consumer laptop chip; they've already got multiple generations mapped out and are treating RTX Spark like an expanding architecture family rather than a single product. He actually showed us that N1X has its own smaller version (N1), so we're looking at sustained evolution here—this isn't a flash-in-the-pan thing, folks.
The vision he painted is absolutely wild: these machines are designed to become what Huang calls "Star Trek computers" and even mentions Star Wars droids as the end state of voice-driven computing interaction! Here's one of my favorite moments from his talk—he compared it directly to Scotty talking with that mouse in *Star Trek IV*, saying we'll walk up to our PCs, say something like "hello, do something," and actually get an intelligent response. He's been collaborating with Microsoft CEO Satya Nadella since around 2023 (about three years now) building toward this goal together. But here's where the really clever part comes in: Huang imagines your R2-D2-like AI working remotely too—not just when you're standing at your desk, but via WhatsApp or other messaging from anywhere! He even gave a gorgeous example of texting his laptop to fix PowerPoint slide 17 while giving a keynote—modifying an image labeled CX9 to CX10 and sending back the finished PDF.
Now here's where Jensen gets practical about local versus cloud computing: he argues you don't want everything in the cloud because if your machine can run things locally, it stays free for daily use (just like why we own our TV or refrigerator instead of renting them), while occasional tools like washer and dryer make sense as services. When his pal Dylan Patel asked "what—am I calling Claude to control my laptop? Are you insane?" Jensen basically agreed—it doesn't even logically work! His entire point is that the files, data, and tools are already local on your machine so of course it should just do things for you without cloud round-trips. What's cool though: The Verge reports these RTX Spark superchips have 128GB of RAM (enough to hold a full 120-billion-parameter AI agent), will scale down to as little as 16GB, and the first-generation models are expected around $3K—so definitely pricey at launch but Huang himself acknowledged that "the power user is the one who has to buy it in this generation." Tom's Guide covered related specs too!
Source: https://www.theverge.com/tech/942588/nvidia-rtx-spark-n2x-n3x-r2-d2-star-trek-star-wars-plan
The vision he painted is absolutely wild: these machines are designed to become what Huang calls "Star Trek computers" and even mentions Star Wars droids as the end state of voice-driven computing interaction! Here's one of my favorite moments from his talk—he compared it directly to Scotty talking with that mouse in *Star Trek IV*, saying we'll walk up to our PCs, say something like "hello, do something," and actually get an intelligent response. He's been collaborating with Microsoft CEO Satya Nadella since around 2023 (about three years now) building toward this goal together. But here's where the really clever part comes in: Huang imagines your R2-D2-like AI working remotely too—not just when you're standing at your desk, but via WhatsApp or other messaging from anywhere! He even gave a gorgeous example of texting his laptop to fix PowerPoint slide 17 while giving a keynote—modifying an image labeled CX9 to CX10 and sending back the finished PDF.
Now here's where Jensen gets practical about local versus cloud computing: he argues you don't want everything in the cloud because if your machine can run things locally, it stays free for daily use (just like why we own our TV or refrigerator instead of renting them), while occasional tools like washer and dryer make sense as services. When his pal Dylan Patel asked "what—am I calling Claude to control my laptop? Are you insane?" Jensen basically agreed—it doesn't even logically work! His entire point is that the files, data, and tools are already local on your machine so of course it should just do things for you without cloud round-trips. What's cool though: The Verge reports these RTX Spark superchips have 128GB of RAM (enough to hold a full 120-billion-parameter AI agent), will scale down to as little as 16GB, and the first-generation models are expected around $3K—so definitely pricey at launch but Huang himself acknowledged that "the power user is the one who has to buy it in this generation." Tom's Guide covered related specs too!
Source: https://www.theverge.com/tech/942588/nvidia-rtx-spark-n2x-n3x-r2-d2-star-trek-star-wars-plan