The profitable deal in Israel and the alarming figure: 5 comments on Nvidia reports

The company reported for the first time revenues of more than 80 billion dollars in one quarter, an annual rate that brings it closer to the threshold of 400 billion dollars in annual revenues, while maintaining a high gross profit (75%) and doubling its free cash flow to almost 50 billion dollars.

Still, one should pay attention to a number of trends that stood out in the chip company’s reports.

1The company’s main source of growth is in Yokneam

At the beginning of 2020, Nvidia completed the purchase of the Israeli Mellanox for close to 7 billion dollars. Last night (Wednesday) Nvidia reported $14.8 billion in revenue from its communications chip operations, another name for Mellanox’s operations currently divided between Yokneam and Tel Aviv.

Nvidia not only returned the price it paid for the Israeli company, which is now seen as significantly lower in relation to the value it provides to the giant, it proves that communications is the growing field in the company: the activity of communications chips jumped by 199% from year to year, and 38% from quarter to quarter.

For comparison, the field of graphic processors grew by only 88% from year to year, and by 17% from quarter to quarter. Despite the extraordinary growth in the field of communication (networking), this is a slow growth rate compared to the previous quarter, which saw a 163% increase from year to year. Nvidia also updated that the number of employees in Israel has already reached 5,900, higher than the official number that was up to now 5,000 employees.

2The new destination: the heart of the Intel empire

In September 2025, Nvidia announced an investment in Intel, then a battered company in the midst of a technological and business crisis, during which one of its stated goals was to integrate Intel’s core processors (CPU) into Nvidia’s systems. Unlike graphics processors (GPUs) that perform the calculation and processing of artificial intelligence, core processors (CPUs) for which Intel is famous manage more general aspects of servers, including artificial intelligence servers, including memory management, running software and load management between the various processors.

The 22% jump in the results of Intel’s server division in the previous quarter illustrated how much the artificial intelligence boom is also tipping the scales in favor of the CPU manufacturers, when for every two graphics processors that are sold, there is also a demand for one core processor. But now Nvidia is signaling that it is no longer content with controlling the GPU market.

At the investor call tonight, Nvidia’s CFO said that the company is on its way to becoming “the world’s leading supplier of core processors”, meaning to overtake Intel in the market share on which it has been weak for the past thirty years. As early as next quarter, Nvidia plans to launch “Vera Robin”, an AI server that will include a combination of its core processors, “Vera”, with the graphics processor “Robin”. For Nvidia, this is a market with the potential of 200 billion dollars, of which it expects revenues of 20 billion dollars already this year.

3The courting of China continues – but the revenues are not coming back

About a week ago, Nvidia CEO Jensen Huang joined US President Donald Trump’s delegation to China at the last minute. Jensen Huang also took advantage of the trip to shoot videos in which he is seen enjoying Chinese street food and said that he “loves China”.

Investors loved the gesture: the stock soared to an all-time high and closed a day later at $235. The hope was that the company would be able to recoup the tens of billions of dollars it used to bring in annually from China before the Trump administration’s export restrictions. According to the reports published tonight, the company still does not expect toSignificant recovery bA powerhouse from the east despite the courtship campaign. The Chinese, apparently, prefer to develop their own Nvidia-compatible chip and forgo the low-end models sold to them from the US.

4Cools the hype around the new competitors

Only a few days after the successful IPO of chip company Cerebras, it seems that Jensen Huang is pouring cold water on its market.

Cerberas is indeed seen as a competitor of Nvidia, but in practice it handles a specific segment of artificial intelligence processing: it addresses a market that requires extremely fast calculation tasks and are intended for a limited number of tasks and users, i.e. for critical areas that require speed of response, such as algo-trading, for example.

Nvidia, which recently purchased a similar company called Groq (Groq) for $20 billion, now claims that “this is a niche market for a while”, and that it will launch a real server called LPX based on Groq’s technology that will be intended for a similar market. “The server will focus on tasks that require low latency and a high token rate, but its ability to serve masses of users (throughput) will be low,” Huang said in the investor call.

5depends more than ever on the giants of the cloud

In the last year, the cloud giants – including Amazon, Meta, Google and Microsoft – have increased their capital equipment expenditures (CAPEX), a code name for graphics processors and AI servers – to an annual rate of 700 billion dollars in total, in order to satisfy the increasing demand of their customers for artificial intelligence.

This is good news for Nvidia, but it increases the relative share of the cloud giants in the revenue pie from 45% in the corresponding quarter to 50% now. This means that half of Nvidia’s revenue depends on how few companies, some of them, Google for example, even develop substitutes for it.