By Daniel Howley

Nvidia (NVDA) is taking aim at Intel (INTC) and AMD (AMD) with the debut of its RTX Spark superchip for Windows laptops. The processor, which includes a Blackwell GPU and Grace CPU, will power laptops from manufacturers including ASUS, Dell (DELL), HP (HPQ), and Microsoft (MSFT) when it lands this fall.

Unveiled during Nvidia’s GTC Taipei event,  the RTX Spark, which is also coming to small desktops, is meant for customers running AI agents, content creators, and, importantly, gamers.

According to the company, the RTX Spark will pack upward of 128GB of memory, a massive amount for any laptop.

Memory serves as a kind of temporary holding area for data that the CPU needs to access quickly. Generally, the more memory, the better the overall performance

Most laptops generally pack 16GB of memory, though higher-end systems, like a top-of-the-line MacBook Pro, can be outfitted with 128GB. But to get that configuration, you’ll have to shell out a whopping $5,099.

Nvidia hasn’t announced pricing for laptops running its new chip, but it did note that the first systems will target the premium market. However, it will also offer less powerful versions of the RTX Spark with less memory for use in lower-priced notebooks.

While Nvidia didn’t provide a look at any of the laptops built using the chip, it did offer hints at styling and features, noting that the notebooks will be roughly 14 millimeters thick, include HD webcams, and have all-day battery life.

Nvidia says RTX Spark laptops will be particularly good at powering AI agents, semiautonomous and fully autonomous digital helpers that perform tasks on your behalf.

Because the RTX Spark is using an Arm-based Grace processor, Nvidia says it’s been working with Microsoft and software developers to ensure that their programs can run on the chip. That’s because the vast majority of software in the world has been built to run on what are called x86 chips from Intel and AMD.

It’s taken years to get developers to either create Arm versions of their apps or build emulators that can translate x86-based programs to run on Arm chips.

And while initial attempts saw less than stellar results, the work has paid off.

Qualcomm’s Arm-based processors already power Windows laptops and can run the vast majority of apps with ease while providing impressive battery life.

But gaming on Arm poses some potential issues.

Nvidia got its start as a gaming company, eventually becoming the largest graphics chipmaker in the world, before it began its work in the AI industry.

And while it’s better known globally for its AI chips, the company’s chips are still highly coveted among gamers. That said, if the RTX Spark can’t run the games players want, it’ll be a hard sell.

Nvidia CEO Jensen Huang introduces the RTX Spark laptop during his keynote speech at Computex 2026 in Taipei on June 1, 2026. (I-Hwa Cheng / AFP via Getty Images)

To address that, Nvidia says it’s working with all of the major game developers to ensure their games and the anti-cheat technology they use to prevent hackers from disrupting games work on RTX Spark-based systems.

What’s more, the company says that the laptops will support its various graphics technologies, including DLSS, which uses AI to improve game performance.

While Nvidia didn’t provide any benchmarks or comparisons to existing laptops, it did say that the RTX Spark’s Blackwell GPU is in roughly the same class as its RTX 5070 laptop GPU.

Interestingly, Nvidia says it doesn’t expect supplies of RTX Spark-based laptops to be limited when they hit the market despite the impact of the ongoing global memory shortage.

Vera production ramping

In addition to debuting the RTX Spark, Nvidia also announced that its Vera data center CPU is now in full production and that Vera-only rack servers will be available this fall.

Nvidia says Anthropic (ANTH.PVT), OpenAI (OPAI.PVT), SpaceXAI (SPAX.PVT), ByteDance, Coreweave (CRWV), and Oracle Cloud Infrastructure (ORCL) are among the customers “exploring” Vera.

CPUs have become an increasingly important part of AI data centers as a result of the increased use of AI Agents.

And while GPUs are still necessary for training and running AI models, the actions AI agents take for you, such as arranging files or browsing the web for you, are powered by CPUs.

That’s made the once also-ran chips far more popular among data center builders and has been a boon for the likes of Intel and AMD, which dominate the CPU data center market.

But with Vera, Nvidia is looking to steal away some of that market share for itself.

Email Daniel Howley at dhowley@yahoofinance.com. Follow him on Twitter at @DanielHowley.