AMD is leaning harder into open-source software as a competitive weapon against Nvidia in the AI chip market, but its stock fell 9% in late trading Tuesday after the company reported earnings. The chipmaker has partnered with OpenAI and Anthropic to promote its open-source ROCm software for programming AI chips, and CEO Lisa Su told analysts the platform is at an "inflection point" with open-source contributions up tenfold in the past year.
Open-source strategy as a cheaper path to speed
Two AMD executives told Business Insider in July that giving away software and hardware designs gives the company an edge over Nvidia and Qualcomm. The core argument: open-source lets AMD tap into thousands of outside developers for free, accelerating development in a race where speed matters more than secrecy.
"The power of the open source community is just unprecedented," said Kirk Saban, corporate vice president of product, software, and solutions. "You can tap into all of those developers, and they're essentially doing free development work for us. They're upstreaming code into our code base, and you're leveraging the best and brightest minds in the industry."
Salil Raje, senior vice president of AMD's adaptive and embedded computing group, called the approach a "central pillar in our strategy." He said AMD's latest chips work best for AI agents partly because developers can improve the software through open-source contributions. "In fact, the reason agentic AI works for our platform better than maybe the competitors' platform is because our software is out in the open now," Raje said.
Earnings: revenue beat, but capex triples
AMD reported revenue of $11.5 billion, beating analysts' expectations, but capital expenditures nearly tripled to $808 million compared with the prior year. The stock sell-off reflects investor concern about rising costs even as demand for AI compute grows. Despite the drop, AMD shares have gained about 193% over the past year, while Nvidia's stock is up roughly 20% in the same period.
Why open source matters for Nvidia's CUDA moat
Nvidia's CUDA software is widely considered a barrier to switching because developers build tools and models on it. AMD's strategy is to make ROCm free and modifiable so that engineers and researchers can adopt it without vendor lock-in. The bet is that openness will attract the developer ecosystem that CUDA currently owns.
For executives and strategy professionals evaluating chip choices for AI workloads, AMD's approach means lower software costs and more flexibility - but also a reliance on community contributions that may not match Nvidia's polished, paid-for tooling. As AMD pushes into physical AI and robotics with the same open-source model, the strategic question is whether community velocity can outrun Nvidia's scale.
Why this matters for Executives and Strategy
For leaders deciding where to place AI infrastructure bets, AMD's open-source bet offers a real alternative to Nvidia's proprietary stack - but it comes with execution risk. The company is spending heavily (capex tripled) while giving away its software, a trade-off that works only if open-source contributions keep accelerating. Executives should watch whether ROCm's developer adoption narrows the gap with CUDA, because that will determine whether AMD becomes a viable second source for AI compute or remains an also-ran. Training teams on the strategic implications of chip openness is one way to prepare - explore AI for Executives & Strategy for frameworks on evaluating AI infrastructure choices.
Your membership also unlocks:
