Nvidia’s Israeli network: The $7 billion bet that built a $30 billion business

When Nvidia announced quarterly results on Wednesday, much of Wall Street’s attention focused on whether the chipmaker could continue to justify its meteoric rise in market value. But beneath the headlines, one part of the business stood out as the quiet powerhouse driving the company’s transformation into a $4 trillion titan: networking.

The division, born out of Nvidia’s 2020 acquisition of Israeli company Mellanox Technologies for $6.9 billion, delivered the most dramatic jump among all of Nvidia’s significant units. Networking revenues surged 46 percent from the previous quarter and nearly doubled year over year, reaching $7.25 billion in the second quarter alone. In other words, in the past quarter alone, the R&D center established through the Mellanox acquisition generated more revenue for Nvidia than the acquisition itself cost. That puts the division on an annual run rate of $25–30 billion, an extraordinary sum for what was once seen as a supporting role to Nvidia’s flagship graphics processors.

Nvidia CEO Jensen Huang and Mellanox founder Eyal Waldman.

Networking now accounts for 16.1% of Nvidia’s total revenue, compared with 11.3% in the first quarter of 2025.

Although the quarterly report labels the segment “Networking,” it could just as easily have said “Israel,” because Nvidia’s communications chip activity is concentrated almost entirely in its R&D center in Yokneam.

In fact, the contribution of the Israeli R&D center to Nvidia’s overall revenue is likely significantly higher than 16.1%. The center may be responsible for close to one-fifth of total revenue. Beyond communications solutions, which constitute the bulk of the activity, the Yokneam teams also develop central processors for data centers (CPUs), systems-on-chip (SoCs) for robotics and automotive, and algorithms for autonomous vehicles. The center also conducts advanced AI research and houses several smaller specialized groups.

New teams in Israel are also working on the software side of AI, including developing language models, based on Nvidia’s acquisitions of Deci and Run:ai. These activities are not broken out separately in the quarterly report, but they are likely to push the Yokneam center’s share of Nvidia’s revenue closer to 20%.

"We have Spectrum-XGS, a giga-scale for connecting multiple data centers, multiple AI factories into a super factory, a gigantic system," said Nvidia CEO Jensen Huang in the earnings calls. "It's the reason why Nvidia dedicate so much in networking. That's the reason why we purchased Mellanox 5.5 years ago. And Spectrum-X, as we mentioned earlier, is now quite a sizable business, and it's only about 1.5 years old. So Spectrum-X is a home run."

Nvidia revealed last week that a technological breakthrough by Nvidia Israel will allow data centers in geographically distant locations to operate as if they were in one place, effectively creating “AI factories” on a massive scale and significantly increasing the maximum computing power available to the industry.

Nvidia’s R&D center in Israel is primarily focused on developing communication solutions for Nvidia’s AI chips. Its two flagship communication products enable ultra-fast connections both between chips within a server and between servers inside the same data center. These capabilities are critical for maximizing the processing power of Nvidia’s AI hardware.

Now, the company is unveiling a new development that extends those capabilities beyond the walls of a single facility. The new technology makes it possible to connect physically distant data centers in much the same way, effectively creating one mega–supercomputer and pushing the boundaries of data center computing power.

According to Nvidia, today’s data centers are already approaching the limits of what a single facility can provide, constrained by physical limitations such as energy supply and chip density. The new platform, Spectrum-XGS, addresses obstacles like long latency, which has until now prevented separate facilities from operating as a unified system.

“The AI industrial revolution is here, and giant-scale AI factories are the essential infrastructure,” said Huang. “With Nvidia Spectrum-XGS Ethernet, we add scale-across to scale-up and scale-out capabilities to link data centers across cities, nations and continents into vast, giga-scale AI super-factories.”

Nvidia Israel’s breakthrough relies on advanced algorithms that enable networks to dynamically adapt to the distance between facilities. “With advanced, auto-adjusted distance congestion control, precision latency management and end-to-end telemetry, Spectrum-XGS Ethernet nearly doubles the performance of the NVIDIA Collective Communications Library, accelerating multi-GPU and multi-node communication to deliver predictable performance across geographically distributed AI clusters. As a result, multiple data centers can operate as a single AI super-factory, fully optimized for long-distance connectivity,” the company said in a press release.

The development highlights the strategic importance of Nvidia’s Israeli R&D hub. Over the next three years, Nvidia’s roadmap includes four major product lines: AI processors, computer processors (CPUs), and two categories of communication chips. Of these, three are being developed under the leadership of the Yokneam-based team. It is likely that the new technology will also be integrated into this roadmap, further cementing the Israeli center’s central role.

This summer, Nvidia announced plans for a massive new campus in northern Israel, with an investment of billions of dollars and spanning 70 to 120 dunams. When completed within five to six years, it will be Israel’s largest high-tech campus and make Nvidia the country’s biggest private-sector employer.

“The synergy between the processor, the brain, and connectivity turned Nvidia from a $93 billion company into the $4 trillion giant it is today,” said Mellanox co-founder Eyal Waldman in a recent interview.

Mellanox, founded by Waldman after earlier stints at Galileo and Marvell, had long specialized in high-performance networking and interconnects. Its InfiniBand technology became essential for linking servers in supercomputers, research clusters, and eventually the vast data centers now used for artificial intelligence.