news

Nvidia has many competitors, both direct and indirect. If you count carefully, half of Silicon Valley is in competition with Nvidia.

However, Nvidia's competitors in the Ethernet field are rarely mentioned. Although the newly launched Spectrum-X is developing rapidly, the strength of switch giant Arista should not be underestimated, especially the tit-for-tat between the two in 1.6T Ethernet products.

Arista also recently announced strong second-quarter results, with revenue growing 16% to $1.69 billion, exceeding analysts' expectations of $1.65 billion. Net profit was $665.4 million, up from $491.9 million in the same period last year. Its stock price rose 3% in after-hours trading. There is no doubt that Arista is one of Nvidia's biggest rivals in the Ethernet field.

However, unlike what most people imagine, Arista is a relatively young company in Silicon Valley, and its founder is not an American, but an out-and-out German. The full name of this German is Andreas Maria Maximilian Freiherr von Mauchenheim genannt Bechtolsheim, or Andy Bechtolsheim for short. Born in 1955, he has lived in the United States for most of his life, but has never tried to obtain American citizenship and has retained his German citizenship.

(Photo from the New York Times)

Becktorsheim's identity is not just as simple as that of the founder of Arista. Today's huge server and workstation market is inseparable from him. Let's take a look at how he, as a German, has gained a foothold in Silicon Valley, USA.

From Bavaria to Silicon Valley

Beckertsheim was born in a lakeside town in Bavaria, Germany. His father is a teacher, his mother is a housewife, and he has an older brother, a younger sister, and a younger brother. In such a peaceful family of six, Beckertsheim showed a strong interest in computers and engineering, and often tinkered with various projects in his basement.

At the age of 16, Beckertsheim began to come into contact with microprocessors and designed an industrial controller based on Intel 8008 for a nearby electronics company. Since he had no access to assembler programs at the time, he programmed it in binary code. For each controller sold by the electronics company, Beckertsheim received a royalty of 100 marks, and his income soon surpassed that of his father.

From today's perspective, Becktorsheim is an absolute child prodigy. In a very restricted environment, without any relevant knowledge, he was able to design industrial controllers on his own. This not only solved his tuition problem, but also encouraged him to continue studying in this industry.

In 1974, Bechtolsheim won the national championship of the "Jugend forscht" physics competition with his research on ultrasonic flow measurement. He later said: "At that moment, my life became interesting. Because the computer equipment of German universities and the entire German computer industry were too backward for themselves."

In 1975, Becktorsheim, who was dissatisfied with the computer environment in Germany, went to the United States with a scholarship. He first obtained a master's degree in computer engineering from Carnegie Mellon University, and then entered Stanford University to pursue a doctorate.

While at Stanford, Bechtolsheim worked on many projects, including as an unpaid consultant at Xerox PARC (Palo Alto Research Center), where he was exposed to early personal computers and envisioned that every scientist could have a large-screen personal computer connected to the Internet, and with the support of his mentor Forest Baskett, he began to work on this idea.

Bechtolsheim eventually designed a Unix workstation called the Sun-1 (short for Stanford University Network). Although it was not the first workstation to be manufactured, it was the first to run on an open system, allowing different types of computers to communicate with each other. The existing technology at the time ran on proprietary closed systems, and Sun's vision was to bring the company into the network age, which reminds us of John Gage's famous slogan - the network is the computer.

But the Sun workstation did not achieve the expected success. Stanford University produced about 10 SUN workstations around 1981-1982. Becktorsheim then sold the workstation design to several computer suppliers, but they all rejected this seemingly advanced product.

Just when Bechtorsheim was frustrated, two other Stanford graduates, Vinod Khosla and Scott McNealy, came to him. The three of them hit it off and later brought in Bill Joy, a Berkeley PhD famous for designing the Unix operating system.

(Image from Google)

In February 1982, Sun Microsystems was officially established, Vinod was appointed president, Scott was appointed director of manufacturing, Andy was vice president of technology, and Bill was responsible for designing workstation software. Sun was an instant success in the university market, generating $8 million in sales in the first two quarters of operation.

Sun was ahead of its time in its vision of network computing power and soon expanded into other verticals outside of the university market, with its first big success being with ComputerVision. ComputerVision wanted a new platform to support its software products and had already signed a contract with another vendor until Sun came along and convinced them that Sun had a better solution. ComputerVision dropped its previous contract in favor of Sun, showing that Sun was a strong competitor in the market. Other companies followed suit, most notably Wall Street financial firms, which became big customers of Sun. Orders poured in, and Sun finally went public in 1986.

Subsequently, Sun experienced even more rapid development and expanded its product range to include Java, Solaris operating system, ZFS, Network File System (NFS) and SPARC. With the popularization of the Internet, a large number of Internet companies were born, and the strong demand for servers continued to raise Sun's revenue and stock price. Before the Internet bubble burst in 2000, Sun achieved revenue of US$18.3 billion, a stock price of US$250 per share, a market value of more than US$200 billion, and 50,000 employees worldwide. Sun reached its peak.

Sun has also left an indelible mark on the technology world. The company is known for its talent and has become a talent training base in Silicon Valley. Many well-known figures in the technology world have worked at Sun and later became CEOs of Google, Yahoo and Motorola. Its most famous software, Java, allows users to write software on any computer and operating system, and it still runs on millions of websites and applications.

But for Bechtolsheim, Sun was just the starting point.

Entrepreneurship and re-entrepreneurship

In 1995, after working at Sun Microsystem for more than ten years, Becktorsheim finally left the company he helped to establish and founded Granite Systems with a large sum of money. The main purpose of the new company was to develop high-speed network switches. One year after its establishment, the company was acquired by the famous Cisco for US$220 million.

After joining, Becktorsheim became the general manager and vice president of Cisco's Gigabit Systems Business Unit. He led the team to develop one of Cisco's successful network equipment series, the Catalyst 4500, which successfully helped Cisco lay the foundation in the Ethernet switching field.

It is not difficult to see from this that Becktorsheim may be a technology fanatic, but he is also an opportunistic businessman.

"I've always been driven by opportunities. We founded Sun around the workstation opportunity (derived from my work at Stanford University)... Later Sun evolved into a server company, which was another great opportunity... In 1995, I saw the opportunity in gigabit networking - so I left Sun to pursue that opportunity. In the end, Cisco acquired me for a huge amount of money," he said in a media interview.

In early 2001, Becktorsheim and Granite Systems partner and Stanford University professor David Cheriton co-founded a new company called Kealia, dedicated to developing advanced server technology using Advanced Micro Devices' Opteron processor. In December 2003, he left Cisco and officially became the head of Kealia.

But he didn't stay independent for long. In February 2004, Sun Microsystems announced the acquisition of Kealia by stock swap. As a result of this acquisition, Beckertsheim returned to Sun again ten years later as senior vice president and chief architect. Kealia's server design concept is supercomputer, but it is aimed at enterprises, and their basic architecture still exists in Sun's product line, especially the Exadata database cluster.

He did not stay long after returning to Sun. In 2005, Beckertsheim and Cheriton co-founded another high-speed network company, Arastra, which was later renamed Arista Networks. In October 2008, Beckertsheim left Sun Microsystems and became chairman and chief development officer of Arista.

At Arista, Bechtolsheim turned to the next wave in networking - 10 Gb/sec Ethernet. A year and a half after he officially joined, Arista Networks launched its first modular switch with 384 10 Gb/sec ports and a variant of Linux Extensible Operating System (EOS) on it, and in March 2011 launched a top-of-rack switch based on Broadcom's Trident+ and Fulcrum Microsystems' Bali switch ASICs.

Gaining an advantage in the network field often means going against the trend. Arista Networks' initial driving force was to use commercial switch chips and focus on developing more flexible network operating systems to attract hyperscale users and cloud builders, while pushing the bandwidth wave as much as possible and encouraging commercial chip manufacturers to speed up innovation and iteration. The company made a fortune from the popularity of 10 Gb/sec, and performed well in the popularity of 40 Gb/sec and 100 Gb/sec. Arista Networks found its own profit point, finding a balance between white box switch suppliers associated with ODMs and large switch OEMs, the latter of which tend to sell products to ordinary enterprises, which often do not have the scalability and software flexibility required by hyperscale enterprises, but they also want to act like hyperscale enterprises and start buying infrastructure in a similar way.

Today, Arista is seizing new opportunities in AI with its bet on Ethernet. Today, Arista is the main competitor of Cisco, which is larger in the computer networking industry. Yes, it is the Cisco where Becktorsheim worked. After many twists and turns, the company he founded has become the biggest rival of his old company. Of course, what is more interesting is that Jayshree Ullal, chairman and CEO of Arista, also came from Cisco, which once caused a lawsuit between the two parties.

(Photo courtesy of Financial Times)

Arista also recently announced strong second-quarter results, with earnings of $2.10 per share after deducting certain costs such as stock compensation, far exceeding Wall Street's target of $1.94 per share. Revenue for the quarter increased 16% to $1.69 billion, exceeding analysts' expectations of $1.65 billion. Net profit was $665.4 million, up from $491.9 million in the same period last year, and its stock price rose 3% in after-hours trading.

The company expects third-quarter revenue of $1.72 billion to $1.75 billion. The upbeat outlook highlights Arista's growing presence in networking, which is benefiting from strong enterprise demand for powerful new generative AI applications that require fast communications infrastructure to run, which is what Arista provides.

Jayshree Ullal, chairman and CEO of Arista, spoke about this need in a blog post, saying that AI training models are particularly dependent on "lossless, high-availability networks to seamlessly connect every GPU in the cluster together and achieve optimal performance." She mentioned that trained AI models also require reliable networks to connect end users and provide fast responses. "As a result, data centers are evolving into new AI centers, and the network is at the center of AI management," Ullal said.

Interestingly, while Arista has benefited from the demand for artificial intelligence, another rival, Juniper Networks Inc., has had difficulty seizing this opportunity. Although it is also increasing its investment in artificial intelligence network products, sales and profits are lower than expected. Due to the development setbacks, the giant Hewlett Packard Enterprise Technology has reached out and intends to acquire Juniper Networks for $14 billion.

From servers to switches, the German genius Becktorsheim has created one storm and revolution after another in Silicon Valley.

Angel Investment

It is worth mentioning that Becktorsheim is not only a tech-savvy engineer and opportunistic entrepreneur, but also a very successful angel investor.

He became active as an investor in the mid-1980s and has participated in the creation of several startups and existing companies. The most valuable investment was in Google, which was still considered a web directory rather than a search engine at the time, but he invested a large sum of money in Google without hesitation.

“We met him on the porch of a Stanford faculty home in Palo Alto,” recalled Google founder Sergey Brin. “We gave him a quick demo. He had to leave quickly, so he said, rather than we go through all the details, I’ll write you a check. It’ll be payable to Google for $100,000.”

His investment in Google's Unique Vision yielded a huge return. A few years after Google went public, $100,000 quickly turned into $10 billion or even more. This investment alone made Becktorsheim a member of the billionaire club.

Of course, it's not just Google. Beckertsheim has made many successful investments in the field of electronic design automation (EDA). His shares in the EDA company Magma Design Automation are worth about $60 million. He is also an early investor in another EDA startup Co-Design Automation, which developed SystemVerilog, which is used in almost all digital hardware designs.

Bechtolsheim follows a simple rule of thumb when deciding whether to invest in an EDA startup. “My whole strategy for companies is if I can’t figure out what they’re doing in a 15-minute call, I’m not interested,” he says. “If something makes sense, I’ll listen more. I invest in companies that make sense to me, that are doing something that needs to be done. If I like the idea, other people probably will too.”

From the late 20th century to the early 21st century, when he continued to invest in EDA companies, Wall Street did not invest much energy in the industry. Investors at the time were busy paying attention to various Internet companies, but Becktorsheim was full of confidence in EDA companies at the forefront of the deep submicron chip revolution.

“Most venture capitalists lost interest in CAD,” he said. “The time to market was too long, and there were too many competitive opportunities in the Internet space. Also, there was some confusion on Wall Street about the CAD revenue model, which led to lower valuations for public CAD companies.”

“I don’t think so. We are about to enter the largest re-tooling of CAD in history, driven by very deep sub-micron technologies. The best next-generation CAD tools will generate significant revenues and company valuations.”

Beckertsheim’s interest in design tools began after he came to the United States as a student in the mid-1970s. In Germany, his engineering tools were pencil and paper, but when he arrived at Stanford University at the age of 19, he fully embraced EDA technology and achieved legendary results.

“I was supposed to work in CAD,” Bechtolsheim said, “but at the time, the only systems were mainframes. I felt it was impossible to run software on a mainframe, and instead we needed a workstation to provide dedicated mainframe computing power to each user. This led me to design what became known as the Sun workstation at Stanford.” This design, in turn, led to “the founding of Sun Microsystems.”

In this small but vibrant field, there are all kinds of startups and many concepts that fail to come to fruition. Beckertsheim typically focuses only on the early stages of the funding cycle, choosing companies that take risks and try to make big strides rather than those that make incremental improvements. "I usually get involved in companies at the first or seed round," he says. "I invest in companies that have the right idea and the right team, and that are doing what needs to be done."

Beckertsheim was interviewed by the media around 2000 and explained his investment layout in the EDA field. He showed great interest in three areas at that time, namely improving verification, accelerating simulation and reducing back-end time.

“We have seen the first results of combining synthesis, place and route tools, and they are amazing,” said Becktorsheim. “This is the breakthrough technology that is really needed. Front-end design can no longer be separated from back-end design.”

Bechtolsheim also pointed out the imbalance between the pace of silicon change and tool advancement. "It's amazing to see the pace of deep submicron technology advancement—chip technology is advancing at an incredible pace," he said. "In my opinion, CAD tools have become the limiting factor. With so much money being invested in fabs and equipment, people are underinvesting in creating and deploying better tools."

In his own investments, “I focus on companies that have the potential to drive a tenfold increase in designer productivity. That’s what’s needed to keep up with future increases in the number of transistors per chip.”

In addition to being an engineer and entrepreneur, his identity as an investor not only allowed Becktorsheim to achieve financial freedom, but also demonstrated his amazing insight in the semiconductor industry. These triple identities ultimately made him one of the most influential figures in Silicon Valley.

The future of workaholics

Thirty or forty years after entering Silicon Valley, Becktorsheim still enjoys his work very much. "For me, work rarely feels like work, but more like fun," said the founding team member of three Silicon Valley companies, the first investor in Google, and a person who has occupied a place on the Forbes billionaire list for more than ten years.

Becktorsheim works because he wants to solve big problems. He succeeds because he never loses sight of the original purpose of his work: satisfying the needs of his customers.

“The great thing about this strategy is that it avoids any internal discussions about what is right or wrong about future demand,” he said. “More importantly, when we develop these products, we know there are customers waiting for them. We know we can sell them and get a good return on our investment.”

For the past dozen years, Bechtolsheim has been working at Arista Networks, where he wears two hats: chief development officer and chairman of the board. The one position he has no interest in taking is CEO. "That's not my skill set," he said simply, explaining that his main focus is working with the engineering team to develop the company's hardware.

Arista CEO Jayshree Ullal, who worked with Becktorsheim after Cisco acquired Granite, praised his vision and dedication as chief development officer. "Andy has repeatedly created disruptive change in Silicon Valley," Ullal said. "Andy is a visionary business leader with a strong sense of pragmatism. He anticipates and pivots to market trends and big markets before they emerge, both in starting companies and investing."

However, Becktorsheim has also encountered some minor troubles recently. In March 2024, the U.S. Securities and Exchange Commission (SEC) accused him of abusing confidential information about Cisco's proposed acquisition of Acacia Communications, and claimed that his colleagues and relatives had made more than $400,000 in profits through illegal options trading. He finally reached a settlement with the SEC on insider trading charges, agreed to pay a civil fine of nearly $1 million, and was banned from serving as an executive or director of a public company for five years.

For Becktorsheim, who has assets worth tens of billions of dollars, this fine is just a drop in the bucket, and being banned from serving as an executive and director does not affect his love for work - Arista confirmed that he is still the chief architect of its network business, and the 68-year-old is still active on the front line of Silicon Valley.

In his book, Is the Rise of Google the End of the Game for Everyone Else?, tech evangelist Jeremy Geelan writes that Bechtolsheim was once asked, "Is the game over?"

His answer was memorable: “Unless someone changes the rules of the game.”