A personal computer made by NVIDIA is on its way. NVIDIA has announced a new PC platform called RTX Spark. Rather than using an Intel or AMD CPU, it is an entirely new computer, with NVIDIA making everything itself, from the CPU and GPU to the motherboard.
These computers will come from ASUS, MSI, Dell, and HP, as well as Microsoft’s top-tier Surface Ultra lineup. Just as Apple struck gold by putting M-series chips into its MacBooks, NVIDIA has built an ecosystem of its own on the Windows side.
Why PCs, and Why Now?
Is NVIDIA making computers really such a big deal? Yes, because this product directly shows that the entire computing paradigm is shifting.
AI is unquestionably the big trend in computing today. One reason Apple’s Mac mini has recently been selling so fast that it has become hard to find is that more people want to run AI agents that act as assistants directly on their computers. Give an instruction, and the computer gathers material, writes emails, and manages information on its own. Naturally, NVIDIA, the company that excels at this AI computation, has stepped forward to take center stage in the PC market.
Another major role for personal computers, even if things have slowed a little lately, is gaming. That is also one of NVIDIA’s core markets. The point is that GPU core counts and memory capacity, rather than CPU-centric performance measures such as gigahertz, have become central to advances in computing. The “Ultra” in Microsoft’s Surface Laptop Ultra is a telling indication of the position NVIDIA’s GPU now holds.
“So NVIDIA Makes CPUs Too?”
Yes. NVIDIA designed its own processor based on the ARM architecture used in smartphones and MacBooks. It runs Microsoft’s ARM version of Windows, so there is no problem using existing apps either. In fact, NVIDIA already sells DGX Spark, a miniature enterprise AI workstation that combines a CPU and GPU in a single chip. DGX Spark runs Ubuntu Linux. This new RTX Spark takes the same chip, installs Windows, and refines it into a PC for everyday users.
You may have concerns about an Arm-based CPU and Windows, particularly if you remember the Surface RT built around an NVIDIA chip years ago. That NVIDIA chip was not based on today’s architecture. It was essentially a smartphone chip running Windows experimentally. The focus was on power consumption rather than computing, so it was positioned no higher than Atom. It was an experimental computer using an NVIDIA chip, rather than a fully fledged NVIDIA platform.
But Windows is now well prepared for Arm, and Arm-based Windows PCs no longer lack performance. In fact, when it comes to overall GPU-led computing, unified memory helps them far surpass most laptops and notebooks.
The GPU’s graphics performance is said to be roughly equivalent to a GeForce RTX 5070. What is more, it uses a unified memory architecture, with the CPU and GPU sharing memory just as they do in Apple silicon. That makes it more versatile than the RTX 5070, whose limited memory is a weakness. It is enough for a Full HD laptop, and the range of AI models it can run should be much broader.
The convenience of getting work done changes completely too. NVIDIA says the age of the keyboard and mouse will eventually fade. We are entering a world where you can open a photo and say, “Clean up the skin and change the expression to a smile,” and the computer opens Photoshop or Premiere on its own and finishes the job. Adobe, CapCut, Blender, and DaVinci are already working with NVIDIA.
NVIDIA’s Revenge on the PC Market
There is also something strangely moving about seeing an NVIDIA computer. CEO Jensen Huang began his career as a CPU developer. He founded NVIDIA with the aim of making a CPU of his own to compete with Intel and AMD. But actually building a processor was technically difficult, and realities such as patents and manufacturing got in the way.
So he turned to graphics cards and GPUs. Even then, though, NVIDIA’s dream of making PCs never faded. Around 2000, the company actually made motherboards for Intel and AMD CPUs, and it also created a gaming laptop platform called Max-Q.
But that journey was anything but easy. Established PC companies were not especially welcoming of NVIDIA’s involvement. In particular, NVIDIA had a fairly bitter dispute with Intel in the late 2000s. As graphics performance became more important, NVIDIA developed a motherboard chipset called nForce that integrated a GPU into the chipset. Today, CPUs include GPUs with basic graphics capabilities, but through the 2000s a graphics card was still essential. NVIDIA wanted to set itself apart in the motherboard market by integrating graphics into the motherboard. But Intel invalidated the license that allowed NVIDIA to make motherboards for Pentium processors, cutting off NVIDIA’s push into the PC market.
Around this time, NVIDIA introduced CUDA, which now drives the AI market, and declared that the center of computing would shift from the CPU to the GPU. Relations between the two companies deteriorated dramatically. NVIDIA ultimately wound down its motherboard business and instead began developing the GPU itself into something like a complete computer. Looking back, that was the turning point that gave rise to the NVIDIA AI empire we know today.
Whatever the eventual outcome, NVIDIA took a serious blow at the time, both to its business and to its pride. It then turned toward the emerging smartphone and mobile markets, but the nature of smartphones brought another formidable obstacle, the modem, and NVIDIA was defeated there too. In the end, though, AI, which turned imagination into reality on the foundation of CUDA, became a major turning point for the company. Now all the companies that disappointed NVIDIA back then are the ones having to come asking for favors.
Ultimately, success in the market is what matters. But for now, RTX Spark looks like such an attractive computing platform that the announcement alone has me checking how much spare money I have in my account.
For decades, NVIDIA was structurally a supporting player in a CPU-centered computing world. But times have finally changed. In a sense, NVIDIA has now reduced the CPU to a supporting role, as the foundation for running the operating system and GPU rather than doing the computing, and made a grand return to the PC market with the GPU doing the real computation.
Look at this a little mischievously, and it almost feels like a patient man’s revenge on Intel and the PC market that left NVIDIA in tears 20 years ago. After a long and winding journey, NVIDIA has realized its original dream, and its RTX Spark-based PCs arrive this fall. I am eager to see whether they can establish AI as a new paradigm in the PC market long dominated by Intel and AMD.
