As regulators clear Nvidia’s $5 billion Intel stake, a chip alliance promises custom AI-era processors, deeper manufacturing ties, and momentum for US’s semiconductor ambitions.
The US Federal Trade Commission has cleared NVIDIA’s plan to take a $5 billion stake in Intel, paving the way for a sweeping collaboration between the two chipmakers. The approval, granted on 18 December, allows the companies to move ahead with a partnership focused on designing and producing custom processors for data centres and personal computers.
The deal also places NVIDIA alongside the US government and SoftBank, both of which recently bought into Intel to bolster domestic semiconductor capacity.
The agreement aims to combine NVIDIA’s high-speed NVLink interconnect, widely used in artificial intelligence (AI) and accelerated computing, with Intel’s CPUs and x86 platforms. Furthermore, Intel will produce custom x86 CPUs for Nvidia’s data centre clients and system-on-chip (SoC) solutions for personal computing customers.
The focus is on creating tailored chips for consumer, enterprise, and hyperscale customers, as computing demands shift rapidly.
The investment still requires customary closing steps, but both companies signalled confidence in the project’s long-term potential, with Huang suggesting it targets markets worth tens of billions of dollars annually.
Nvidia CEO Jensen Huang rejected suggestions that geopolitical tensions or regulatory pressure prompted the partnership. Reports have highlighted China’s restrictions on purchases of NVIDIA’s AI chips, but the NVIDIA chief said the collaboration was not driven by Washington, Beijing or any coordinated policy effort. He added that the two firms had been working together quietly for almost a year.
Intel CEO Lip-Bu Tan described the collaboration as transformative, though manufacturing specifics remain uncertain. Intel has previously indicated it may outsource some advanced production if needed, including to TSMC.
Executives offered few details on where the new chips will be built, but stressed that the immediate priority is executing the collaboration itself.
