Hardware engineers got a pay bump in 2025

It paid to be a hardware engineer in 2025, according to data from Levels.fyi—largely because of the desire among companies for AI-powered operations.

The increasing use of AI and automation requires professionals who can support a growing demand for the custom GPUs, chips, data centers, and other hardware connected to IT infrastructure.

“There’s a huge demand for AI/ML infrastructure,” Levels.fyi co-founder Zuhayeer Musa said, adding that the hardware profession—which has perhaps taken a back seat to software roles in recent years—is seeing a “resurgence.”

A “Wrapped”-style 2025 report from Levels.fyi found a 15% year over year pay growth for hardware engineers, along with a median yearly salary of $225,000. The data also revealed a notable 9.6% year over year pay bump for software engineering managers, as well as a 4.6% increase for the product manager.

Some companies looking for a hardware engineer want expertise in AI server hardware built by companies like Nvidia and AMD. Just this month, Nvidia teamed with Siemens to integrate AI into industrial workflows on the factory floor. Meanwhile, AMD announced a partnership in October 2025 to power OpenAI’s infrastructure with its AMD Instinct GPUs.

A representative job post for a hardware engineer asks for expertise in tasks like:

• Designing AI server infrastructure (including GPU clusters) for deep-learning workloads
• Configuring switches and understanding power-management and cooling needs for data-center environments
• Evaluating “next-generation” GPUs, TPUs, and custom AI accelerators
• Setting up racks and electrical panels to support AI computing requirements

“There are trillions of dollars of [AI] infrastructure that needs to be built out,” Nvidia CEO Jensen Huang told a crowd at Davos last week.

Aside from AI upskilling, Greg Fuller, VP of Skillsoft Codecademy Enterprise, recommends today’s hardware engineers review business-impacting code in their environments, and to learn more about programming languages like Rust, which compiles directly to machine code, leading to an efficiency boost for compute-heavy tasks like AI. (“That’s one of our trending languages that really fits into the hardware space as well,” Fuller said.)

As software giants like Google add TPUs and and Meta tests out its own in-house AI chips, Musa believes the engineers need to be ready to handle the further convergence of software and hardware.

“I think one of the skills that we’ll continue to see going into the new year is you’ll have these hybrid teams that are working on optimizing hardware, but also optimizing the software around it,” Musa said. “I think one of the strongest ways to kind of position yourself is to not just be a great hardware engineer, but also be a great software engineer, and noting that they are going to be more intertwined going forward.”