Nvidia has spent the last two years selling picks and shovels for the AI gold rush. Now it's trying to sell the gold miners their offices, too.
Over the weekend, Nvidia and Microsoft unveiled RTX Spark, a new class of AI-powered Windows PCs that can run advanced AI models directly on users' devices.
Nvidia says the systems can deliver up to a petaflop of AI performance, enough horsepower to make your current laptop look like it still pays for AOL.
Wall Street immediately understood what was happening. The announcement wasn't just about a faster PC. It was about Nvidia expanding its territory.
New Frontier
Until now, Nvidia's AI empire has largely lived inside massive data centers owned by Microsoft, Amazon, Google, and Meta. Every time ChatGPT answers a question or Claude writes a paragraph, there's a decent chance Nvidia silicon is doing the heavy lifting somewhere in the background. RTX Spark changes the equation.
Instead of sending workloads to the cloud, users could increasingly run AI models locally. That's good news for Microsoft, which wants Windows to become the operating system of choice for AI agents. It's also good news for Arm, whose chip architecture powers the new devices. The market reaction reflected exactly that.
Shares of Arm jumped while investors dumped Intel and AMD, whose dominance in personal computing suddenly looks a little less secure. Qualcomm also found itself in the crosshairs as Nvidia crashed the AI-PC party it had hoped to help host.
Whether consumers actually rush to buy AI PCs remains an open question. The tech industry has spent the last year insisting AI PCs are the future, while most consumers have responded by asking whether their current laptop still opens Excel. That's a problem. Because revolutionary products only become revolutionary when people actually want them.
The Bottom Line
Nvidia isn't just trying to own the AI infrastructure powering the future. It's trying to own the device sitting on your desk, too. If successful, the company could extend its AI dominance from the data center all the way to the end user and give Intel, AMD, and Qualcomm something new to worry about.