Arista is a key AI network-layer play: open Ethernet switching, supply-constrained demand, cloud customers diversifying from proprietary options. FY2026 revenue guided to $12.6B and AI revenue to ~$3.6B, after Q2 revenue topped $3B, up ~38% y/y. Risks: customer concentration, supply constraints. Worth an AI watchlist.

The last stock I wanted to talk about in today's video is Arista Networks. If you are not familiar with Arista Networks , this company makes network switches and software that transfer data between servers within data centers.

That is, it provides this type of high-speed network that connects thousands of individual servers and, of course, increasingly, AI acceleration chips , into a single computing system.

Arista Networks does not manufacture chips or manage cloud infrastructure; rather, it manufactures the network layer to which other computing and storage devices connect, and it is built on its own proprietary software platform that runs on all of its products.

This means that customers enjoy a unified and consistent operating environment regardless of the model of converter they use. Therefore, if the network connecting them is slow or unreliable, these expensive chips may remain idle waiting for data instead of being used for computing , meaning that the network layer directly determines the value a company gets from the graphics processing units it has already paid for .

This is what makes Arista Networks a major player in this field.

Training a massive artificial intelligence model requires not just powerful chips, but thousands of chips to communicate with each other almost instantly . The networking technology that connects them has become one of the most significant obstacles we are witnessing as artificial intelligence groups expand.

Therefore , Arista is striving to be one of the key players in solving this problem.

Arista has established itself in the Ethernet- based networking market as a more open and flexible alternative to other proprietary interconnection technologies, such as Nvidia’s InfiniBand technology . This approach has proven successful.

In recent earnings reports, management stated that demand for AI networks has reached a point of supply bottleneck , and that Arista is now experiencing supply constraints involving chips, optics, and components.

It is worth noting that competitors, such as Nvidia's InfiniBand technology, restrict customers to Nvidia's integrated system.

So, Arista's offer to cloud computing giants is, in short: " You can build an AI suite that performs as well using open Ethernet standards, without being restricted to a proprietary networking tool from a single vendor."

This is one of the main reasons that enabled Arista to strike deals with companies like Meta Platforms and Microsoft, which theoretically could have relied entirely on a single supplier's solutions.

But as these obstacles have become increasingly apparent, it is also clear that giant cloud computing companies are seeking to diversify the suppliers they collaborate with. This was, of course, a crucial factor in the continued growth of Arista Networks.

In the second quarter of its fiscal year 2026, the company announced last August that its revenues had exceeded $3 billion for the first time ever, an increase of approximately 38% year-on-year.

Management raised its 2026 forecast to $12.6 billion , representing growth of nearly 40% compared to fiscal year 2025.

The company also raised its AI revenue target to approximately $3.6 billion for this year. The company has so far signed contracts with more than 100 customers for its Ethernet switching platform in the field of artificial intelligence technology .

In comparison to the past few years, specifically 2024, if you have been following the company for a while, you will find that the number of early users was very limited. They have achieved tremendous growth in this area over the past two years.

It is estimated that Arista Networks’ total target market, with its Ethernet switch platform and all AI customers , could reach more than $105 billion .

Arista now works directly with several giant AI service providers . We are witnessing a shift towards Ethernet-based AI networks , which is beneficial to the company. Arista also introduced next-generation AI network platforms with speeds of 1.6 terabits per second, as well as liquid-cooled switching options.

These platforms are specifically designed to train the largest AI training pools , placing them at the forefront of the next wave of pool scaling, meaning they not only serve current deployments but also look to the future and growth opportunities.

One of the things I like about Arista Networks is that, in addition to facing a real challenge in building this massive AI infrastructure, they have also built a solid reputation in the hardware and software field over the years.

Arista achieved a net customer satisfaction rating of 89, which is among the highest rates among enterprise technology companies. This means that, as many major players seek to address these obstacles in AI development, Arista is a reliable and preferred option.

Of course, the net customer satisfaction rate is a very strong indicator, especially for enterprise hardware companies. Brands strive to achieve net customer satisfaction rates like these.

This has been a key part of Arista's career for a long time.

As for AI revenues , there is a risk of customer concentration, which is worth noting. A relatively small number of customers from large data centers constitute a large and growing share of the artificial intelligence business.

These customers are also the key players driving most of the AI spending we are currently seeing. But this is noteworthy, and of course, there are always supply chain constraints that may hinder growth from quarter to quarter.

But Arista has become one of the most important, if not the least well-known, companies in the field of artificial intelligence infrastructure, specifically the network layer that must be operational for the training suite to function .

I believe the danger lies not in the existence of a real bottleneck in AI networks , but in the extent to which this opportunity is concentrated among a number of large clients.

I think this will remain a major bottleneck in the coming years, and companies like Arista Networks can help address it.

It is an efficiently managed company, and I believe it deserves to be added to your watchlist if you are growing your portfolio in artificial intelligence stocks. Yes, and if we look at the last 12 months, we find that the revenues achieved have slightly exceeded $10 billion , and if we look at the market expectations for this company in two and a half years, it is expected to double.

These revenues are expected to double . So, there is undoubtedly a large demand.

Well, there is another area that is extremely important, and a key part of this AI story, and that is Arista Networks.