Nvidia’s $6.9 billion Mellanox bet in 2020 is paying off, analysts say

In 2020, Nvidia made a sizable purchase that many investors had never fully connected to its core business. The company agreed to spend $6.9 billion on Mellanox, an Israeli company best known for making networking switches and adapters, according to a report published by Yahoo Finance. At the time, the deal drew limited attention compared with Nvidia’s faster-growing visibility in graphics and computing hardware.

Nvidia has since positioned itself as a central supplier for AI infrastructure, where massive computing clusters require more than just high-end GPUs. Between those chips, systems need extremely fast and reliable connectivity to move data during training and inference workloads. The Mellanox products Nvidia acquired were built for that type of requirement, the report notes, which is part of why the earlier investment is now being viewed in a new light.

The “side bet” framing captures how the deal initially landed. A company focused on specialized data-center networking was not as widely discussed in the mainstream as Nvidia’s flagship chip architectures and developer ecosystem. Yet the market’s understanding has shifted as AI deployments expanded from isolated models to large-scale data-center operations, where networking becomes a practical bottleneck or an enabler depending on performance.

Mellanox’s role, as described in the report, was not a generic hardware contribution. It manufactured the kind of infrastructure components that allow high-performance systems to interconnect at scale, including networking switches and adapters. Those building blocks matter when data-center operators assemble racks of processors and want the cluster to behave like a single, tightly connected computing pool rather than a collection of loosely linked machines.

For Nvidia, the acquisition also fit a broader strategy: building a more complete platform for AI computing rather than selling a single chip. When the company sells a full stack, from compute to the interconnect that ties compute nodes together, customers can deploy systems with fewer integration steps and less uncertainty about performance. Even so, Nvidia has not publicly laid out, in the Yahoo Finance report, a simple scoreboard that ties the Mellanox purchase to a specific line item today.

The pay-off thesis is inherently backward-looking, and the report provides more narrative than accounting. It does not, in the material provided here, break out how much revenue or margin growth can be directly attributed to Mellanox’s technology, nor does it detail any internal metrics such as deal-driven share gains in networking components. As a result, the timing and magnitude of the benefits are best understood as a “platform relevance” argument rather than a fully quantified impact statement.

Still, the broader implication is clear: as AI workloads have grown more compute-intensive, the supporting infrastructure around those workloads has become more strategic. The market is increasingly rewarding companies that can supply not only processing power but also the interconnect that helps those processors work efficiently as a system.

What to watch next is whether Nvidia continues to expand its networking presence alongside its AI and data-center roadmap, and whether more detailed disclosures emerge from the company on how its networking capabilities contribute to product demand. Investors will likely look for additional indicates about adoption and customer deployments that depend on the acquired technology.