Jayshree Ullal had already done the sensible thing. She had built a career at Cisco, risen to senior vice president and taken responsibility for a business measured in billions. Then, in 2008, she left. The company she joined had no revenue. For an executive who knew exactly how much work went into making a networking business function, zero was a particularly informed choice.
The decision is easier to admire from a distance. Arista Networks went public in June 2014 and reported $9.006 billion in revenue for 2025. Those results give the opening chapter a reassuring ending. But they can also obscure what made the move interesting: Ullal had considerable experience, considerable status and an appetite for another beginning. She wanted to use what she knew somewhere she could help build the business around it.
Today, as Arista’s CEO and chairperson, she works in an industry where artificial intelligence has made the connections between computers a matter of enormous commercial interest. Her route there began with an interest in science, then widened to include people. For someone trained to design chips, she developed a useful fascination with the customers at the other end.
A useful interest in other people
Born in London and raised in New Delhi, Ullal came to the United States for higher education. At San Francisco State, where she enrolled in 1977, she studied electrical engineering. The degree followed her strengths in mathematics and science. It also put her close to an industry inventing new things at a pace that made a classroom feel unusually near the action.
There were less glamorous encounters with computing, too. She has recalled driving to Silicon Valley to obtain flawed chip samples for programming work. Good chips were for paying customers; a student could make use of the leftovers. Once, she dropped a stack of programming cards and faced having to do the work again. Computing offered many opportunities to learn precision, some of them administered by gravity.
She graduated in 1981 and entered semiconductor engineering, working at Fairchild and AMD. Her education continued with a master’s degree in engineering management at Santa Clara University in 1986. The pairing would prove apt: an understanding of how technology works, alongside an interest in getting people and businesses to make something useful of it.
“I liked people and communication, which I consider to be anti-engineering traits.”
Jayshree Ullal, 2006
The joke contains a fairly serious career decision. As a chip designer, Ullal found that she liked working with customers and solving their problems. A technical question could arrive wrapped in a commercial one. Understanding the circuitry was useful; understanding what somebody needed the circuitry to accomplish opened another line of work.
That shift makes her later career less mysterious. Marketing and engineering often get assigned opposing temperaments. Ullal had reasons to be interested in both. She could follow a product into the details and follow a conversation beyond the product. The industry would give her plenty of opportunities to practice the combination.
Fifteen years inside the machine
After semiconductor work and a role at networking company Ungermann-Bass, Ullal joined Crescendo Communications. She was its vice president of marketing when Cisco acquired it in 1993. Crescendo was Cisco’s first acquisition, and its switching technology helped form the basis of the Catalyst business. An acquisition that might occupy a line on somebody’s résumé became the entrance to fifteen years of her working life.
Ullal had expected to stay for two. At Cisco, the work kept changing enough to make staying interesting. The company was expanding, and she moved through leadership roles as switching became a substantial business. By 2005, she was senior vice president of the Data Center, Switching and Security Technology group. The scale had changed rather considerably from the semiconductor bench.
There is a particular education in growing within a company that is itself growing. Teams need integrating. Products need customers. A promising technology needs a route from the engineer’s demonstration to somebody’s operating budget. Ullal’s Cisco years gave her experience of those transitions from inside a business that could fund them at scale.
Eventually, she asked whether she wanted to retire there. The answer sent her elsewhere. She took the summer of 2008 off, considered different fields, including clean technology, and found her way back to networking. Andy Bechtolsheim, whom she knew from Cisco, was part of the attraction at Arista. Starting again would involve people she already understood.
A company small enough to start again
Arista announced her appointment as president and CEO on October 23, 2008. Bechtolsheim became chief development officer and chairman. The announcement put an experienced networking executive beside a founder with his own history of building technology companies. Kenneth Duda and David Cheriton were also central figures in Arista’s founding story. Ullal joined a technical undertaking that needed to become a commercial company.
Her responsibilities at Cisco had included a $10 billion business. Arista’s sales, she later recalled, were being counted in increments of $10,000. That is a change in scale large enough to alter the meaning of a good day. A modest order at a large corporation can be a consequential event inside a young business.
What Arista had was a specific proposition. Its Extensible Operating System, EOS, put software architecture at the center of the networking equipment. The company’s 2008 announcement already emphasized that software, including its modular design and ability to support upgrades while systems remained in service. Ullal was joining a business with an engineering argument to make.
EOS remains central to Arista’s approach. A common software image runs across its switching family, with a programmable, Linux-based system underneath. For a customer operating a large network, consistency has a practical attraction: fewer differences to accommodate as the equipment grows. A switch may be a physical box, but its behavior is a software question.
That suited the work Ullal had learned to do. Engineers could explain the architecture; customers could explain the operational problem. Building the business required a convincing relationship between those explanations. The hardware had to perform, and the customer had to see why this particular design would make a difficult job easier.
The numbers catch up
In June 2014, Arista listed on the New York Stock Exchange under ANET. A public listing changed the company’s financial circumstances and the scrutiny applied to it. It also supplied a visible marker for a quieter process: a young networking business had found enough customers, and enough confidence in its future, to reach the public market.
By the end of 2025, Arista reported cumulative shipments of 150 million ports. Full-year revenue rose 28.6% to $9.006 billion. These are company achievements, produced by many people over many years. They also give the size of the organization Ullal has led since its early commercial days a less abstract description.
The next reported milestone arrived in August 2026: second-quarter revenue of $3.036 billion, the first time Arista had crossed $3 billion in a quarter. Revenue was 37.7% above the same period a year earlier. The annual and quarterly figures measure different periods, but together they show a business still growing after its initial cloud-networking opportunity.
Recognition followed the operating results. Ullal received EY entrepreneur recognition in 2015, appeared among Barron’s selected CEOs in 2018 and Fortune’s business people in 2019, and was named Global Indian of the Year by The Economic Times in 2023. In June 2020, she joined Snowflake’s board, taking her experience into another company built around enterprise data.
The next distance to cross
AI has brought a fresh version of an old networking problem: useful computing depends on the connections between machines. More computing capacity does not, by itself, settle how information travels between those machines. Delays, congestion and failures still have to be dealt with. The scale and coordination of the work make those details increasingly consequential.
In September 2026, Ullal and Brendan Gibbs wrote about networks connecting AI infrastructure across locations. Their argument starts with physical limits: space and power can constrain how much computing fits in one place. Connecting sites creates another set of engineering demands, including predictable performance, security and recovery when something goes wrong.
This is a concrete expression of her current ambitions for Arista. It also connects to the commercial judgment behind her 2008 move. A change in computing creates a problem for the network. A company has to decide whether it understands that problem well enough to build around it. Ullal has now faced versions of that decision across more than one industry cycle.
Back in the classroom, facing the other way
In May 2023, Ullal returned to San Francisco State as its commencement speaker. She described arriving as a shy student in a field where women were scarce, and remembered the support of teachers and classmates. She even found a practical disadvantage in being so conspicuous: skipping class was difficult. A future chief executive had once faced a rather efficient attendance system.
Santa Clara University brought her back in June 2025, awarding her an honorary Doctor of Engineering as she addressed its graduating class. Nearly four decades had passed since her master’s degree. She spoke about a career made through experiments, learning and contributions, and about the value of the people around you. The advice fits a career whose important turns involved both technical decisions and human connections.
Seen from those campuses, the progression is unusually legible: a student interested in science, an engineer interested in customers, an executive willing to exchange a large organization for a small one. Each step used something from the previous step without requiring her to remain there. In 2008, zero gave Ullal room to try again. The experience came with her.