Today, NVIDIA announced the private placement purchase of Intel Corporation shares worth $5 billion.
The graphics chipmaker initially revealed its investment plans in September, stating it would acquire approximately 214.7 million shares of Intel—representing about 4.4% ownership—at a price of $23.28 per share. This price was set 5% below the closing market value on the day prior to the announcement.
This strategic investment is part of a broader collaboration between the two tech giants, who will also co-develop multiple engineering initiatives. The first major project involves Intel designing central processing units optimized for NVIDIA’s graphics processing units, centered around NVLink—a high-speed interconnect technology developed by NVIDIA to enhance chip-to-chip communication.
Most hardware manufacturers rely on PCIe (Peripheral Component Interconnect Express) to manage data transfer between server processors. The latest iteration, PCIe 5.0, supports up to 128 gigabits per second. In contrast, NVIDIA's NVLink can achieve speeds as high as 1.8 terabits per second—more than ten times faster. Intel’s upcoming server CPUs are expected to leverage this technology for significantly accelerated data exchange with GPUs.
Collaboration also extends into the consumer market. Intel intends to integrate chiplets derived from NVIDIA’s discrete desktop CPUs into system-on-chip (SoC) designs. Chiplets refer to compact computing modules embedded within larger processor architectures to boost performance and efficiency.
Reports indicate that NVIDIA had previously considered using Intel’s advanced Intel 18A process node for manufacturing certain chips. This fabrication node marks one of Intel’s most significant foundry upgrades in recent years, featuring a novel transistor architecture and optimizations that mitigate power fluctuations within the chip—issues that could otherwise degrade performance if unaddressed.
Last week, Reuters reported that NVIDIA has halted a series of tests aimed at evaluating whether the Intel 18A process is suitable for GPU production. The pause may be linked to delays in rolling out the node at scale. Yield rates for Intel 18A, a key metric for production efficiency, are not expected to reach industry-standard levels until 2027. Apple is also believed to be assessing the same manufacturing process. Last month, prominent analyst Ming-Chi Kuo stated that Intel’s chances of securing a contract to produce chips for the iPhone maker have "significantly increased." Such a deal could involve manufacturing M-series processors used in devices like the iPad Pro and entry-level MacBook Air.
