Nvidia's announcement of a new PC chip set markets ablaze on Monday. Chip stocks, led by Intel and Qualcomm, fell sharply, while PC stocks, led by Dell Technologies and HP Inc. jumped. Everyone should take a deep breath.
A new kind of PC, even one armed with an Nvidia chip, will be an uphill climb. Amid a memory shortage, it could hardly be a worse moment to be launching a new PC platform. The history is tough too: a 15-year project of expanding Windows PCs beyond Intel and Advanced Micro Devices has seen little payoff.
Steep memory price hikes are hurting sales of all kinds of consumer electronics, nowhere more so than in PCs. Market research firms and PC makers alike expect unit sales to shrink by almost 20% in the second half of this year, with most of the damage coming in low-price tiers. Among other mitigation efforts, manufacturers are raising prices.
Ironically, it's Nvidia, the data-center chip maker, that's responsible for much of the shortage in the first place. This year, just five companies may spend over three-quarters of a trillion dollars on AI data centers, and much of that goes for Nvidia chips. Each of Nvidia's coming Vera Rubin servers essentially uses the memory equivalent of 14,500 MacBook Neos.
Nvidia is likely targeting the high end of the PC market, which is less affected by the memory price hikes. The bigger problem for Nvidia is years of false starts when it comes to running Microsoft Windows on anything but x86 chips.
Since IBM chose an Intel CPU for its first PC in 1980, chips based around the x86 architecture have run Microsoft DOS and Windows PCs. AMD also makes x86 PC chips and has about a quarter of the market now.
Microsoft has tried to break its x86 reliance, with little success. Chips using architecture from Arm Holdings now dominate smartphones and Apple MacBooks, but they have been notably absent from Windows PCs; Microsoft, Qualcomm, and Nvidia have been trying to change that since 2011, and it's still a work in progress.
Meanwhile, Apple leapfrogged them. It had been making its own Arm chips for all its other devices since 2010, and in 2020 it debuted more powerful ones for Macs. The almost seamless transition from Intel chips was shepherded by John Ternus, who will become Apple CEO in September.
Using Arm chips, Macs had a big advantage over Windows laptops, with slim, light designs, fast performance, quiet cooling, and all-day battery life. The new $600 MacBook Neo ($500 for students), built around one of Apple's iPhone chips, brings a level of sophistication to that price-point that PCs can't match.
The pressure from Apple added more urgency to the Windows on Arm project, but efforts suffered from both hardware and software issues. A watershed moment in the project came in 2024, when PC makers like Dell, HP, and Samsung Electronics began releasing laptops built around Qualcomm's new Snapdragon X PC platform. Common apps like the Microsoft Office productivity suite and Google Chrome run smoothly, and PCs once again rival Macs in battery life.
Still, PC buyers aren't jumping in like they did with the Arm-powered Macs. Qualcomm is targeting $4 billion in PC-chip sales in 2029, a modest goal considering that Intel and AMD had combined 2025 PC revenue of $43 billion.
Now it's Nvidia's turn to try to make Windows on Arm happen. But in this area, Nvidia looks more like Qualcomm than Apple. Apple deftly made the transition to Arm Macs because it controlled every part of the product: The chip, the device, the operating system, many of the apps, and the developer tools. Apple also had the experience from a similar transition to Intel chips 15 years earlier. And most crucially, Apple stopped selling Intel Macs.
Nvidia has a similar level of control in data centers. In PCs, it will be dependent on other companies putting pieces together. PC makers like Dell and HP will give it a go with Nvidia chips, but they haven't forgotten past failures with other Arm products. Don't expect them to throw everything at Nvidia PCs.
As for Nvidia stock, even PC success may not be enough. Shares rose 6.3% on Monday, boosted by PC excitement. It's an overreach given that the entire PC market is a fraction of Nvidia's expected $364 billion in data-center sales this year. If Nvidia takes 50% of the x86 PC market, it would only raise next year's projected sales by 4%.