Upgradable Laptop GPUs Have Arrived

Framework and Nvidia develop a swappable RTX 5070 graphics module

3 min read

Matthew S. Smith is a contributing editor for IEEE Spectrum and the former lead reviews editor at Digital Trends.

A person easily swapping out the graphics processing unit inside their laptop.

Users can now swap out or upgrade the GPU in a Framework 16 laptop.

Source images: Framework

The GPU is among the most important silicon in a modern laptop. It’s often tapped for demanding tasks, including 3D graphics rendering, video encoding, and AI workloads. Yet despite its importance, swapping the GPU in a modern laptop is impossible: It’s soldered to a mainboard, and upgrades aren’t available.

Until now.

Framework, a small laptop maker focused on repairability, released a new upgradable graphics module late last month. It includes the Nvidia RTX 5070 GPU, fits in all existing Framework 16 laptops, and can be installed with a single screwdriver.

“Upgradable graphics in a laptop is basically the holy grail,” says Nirav Patel, the CEO of Framework. “In the desktop world, graphics upgrades are among the most common behaviors […] it’s one of the core reasons to even buy or build a desktop PC. So we knew that was something we wanted to do.”

Framework’s Swappable Graphics Module

Upgrading laptop graphics isn’t entirely new. About a decade ago, a handful of manufacturers sold laptops compatible with Mobile PCIe Express (MXM), a compact PCI Express standard. But MXM never caught on. Card makers often ignored the specifications. MXM graphics modules designed for one laptop often didn’t fit another (even among laptops sold by the same company), and when they did fit, firmware complications meant it wasn’t guaranteed to work.

Framework took a different approach. Instead of developing a standard for the industry, the company focused on building a swappable graphics module for a single laptop: the Framework 16. The module attaches to the laptop’s rear and connects via a custom interposer secured with screws, making it possible to swap the module in a few minutes. The first module launched in 2024 with AMD’s Radeon RX 7700S.

It was the only GPU available, however, which meant it didn’t yet deliver upgradable graphics. Framework promised future upgrades, but users were skeptical. After all, MXM—a standard backed by many larger laptop makers—made similar promises and failed to deliver.

Framework’s introduction of a new graphics module with Nvidia’s RTX 5070 fulfills its promise. The new module slots right into the Framework 16’s existing graphics bay, connects to the same custom interposer, and is secured with the same screws. Both graphics modules weigh 0.3 kilograms and look nearly identical. Even the internals are difficult to tell apart, though a close inspection will reveal subtle differences in the heat pipes and cooling fans.

Patel says Framework and Nvidia began work on the RTX 5070 graphics module in 2024, shortly after the original graphics module was released, and partnered closely to make the upgrade a reality.

“We were essentially the only laptop maker not shipping Nvidia at that point, so of course, they want to win every customer they can,” says Patel. “But also, they know the Framework 16 is unique and compelling, and solves a problem others haven’t been able to solve.”

Framework picked Nvidia’s RTX 5070 for its second graphics module because the company felt it filled the power and performance “sw­eet spot” for an upgradable GPU. Introduced earlier this year, the RTX 5070 boasts Nvidia’s new Blackwell architecture and provides a considerable upgrade over the older AMD RX 7700S. But the RTX 5070 is also similar to the RX 7700S in thermals and physical size, making it a good fit for the existing graphics module design.

While it’s fair to say the Nvidia GPU has a performance edge over the AMD option, the existing AMD graphics module remains relevant and will continue to be sold. AMD provides open-source drivers and is sometimes preferred in the Linux community, which is a key demographic for Framework—though Patel says Nvidia’s Linux driver support has “matured a lot” and shouldn’t be a problem “in the [distributions] that we recommend people use.”

Introducing the 240-Watt USB-C Power Adapter

The Nvidia RTX 5070 graphics module was the highlight of Framework’s recent announcement, but it came with a companion that also deserves the spotlight: a 240-watt USB-C power adapter, the first of its kind for a consumer laptop.

Framework previously shipped the Framework 16 with an 180-W power adapter, which was already more capable than most (USB-C laptop adapters are typically 65 to 100 W). The 180-W adapter “is basically enough for most workloads,” Patel says, but in high-demand scenarios, power draw could rise higher.

The 240-W adapter also brings new functionality. While the original AMD graphics module supported only display output over the included USB-C port, the RTX 5070 module can accept power input as well. Firmware work was required to make the port’s power delivery, display output, and data modes work seamlessly through the interposer that connects the graphics module to the laptop.

So why haven’t other laptop makers introduced 240-W USB-C adapters for their own high-performance laptops?

Part of the answer is supply-chain issues. The company moved production from China to Thailand to avoid potential tariff issues, though it continued working with the same manufacturing partner at a different facility. The company’s laptop assembly remains in Taiwan.

Patel also says that designing a power-delivery system capable of handling the 240-W adapter required extra engineering effort. “For the first generation of Framework Laptop 16, the silicon didn’t exist for us to be able to create the [240-W] power adapter yet,” he says. “As soon as the technology was available, we went ahead and did that.”

The Conversation (0)