Sanjay Uppal and the Question That Outlived Three Acquisitions

The hotel guests had the better deal. They could open their laptops and use inexpensive broadband, while the people running the business wrestled with the machinery of corporate networking. During VeloCloud’s early market research, a hotel-chain IT chief asked why the connection serving the guests could not also carry the company’s applications. It was an awkward question for an industry accustomed to selling elaborate answers. For Sanjay Uppal and his fellow founders, it was a useful one.

The complaint gave their work a practical target: make ordinary connections usable for business, with the security, reliability and control that a company required. The hotel manager was asking for less fuss. Uppal would spend the next decade turning that request into a company, then carrying it through a succession of much larger companies. Somewhere along the way, the industry acquired another acronym: SD-WAN.

Uppal’s professional biography supplies a less formal credential. He likes making gadgets, gardening with his family and watching cult films. His X biography declares a belief that life’s lessons can be learned from The Big Lebowski. That is a generous educational burden to place on a film involving a disputed rug. It does, however, make an unexpectedly suitable companion to a career spent getting complicated things to connect.

Three founders, one stubborn question

VeloCloud was founded in 2012 by Uppal, Steve Woo and Ajit Mayya, working with the venture co-creation firm The Fabric. Woo came from product management at Aerohive; Mayya had been a VMware engineering leader. Uppal brought a career crossing engineering, business development and executive management. This was a group with substantial experience of the equipment and organizations it hoped to change.

The Fabric’s early account describes a deliberate process of testing the idea with prospective customers and partners. Broadband was becoming more capable. Applications were moving into the cloud. Yet the network linking a company’s scattered offices still presented a problem. The hotel manager’s question helped turn those broad trends into a recognizable buying decision.

In July 2014, VeloCloud made its cloud-delivered WAN service generally available. It had raised $21 million, and customers could pay a monthly subscription. Uppal explained that the service could combine familiar connections, including cable, DSL and LTE, and steer applications toward an appropriate path. The proposition had an appealing lack of ceremony: use available links more intelligently, and spare the branch office some of its infrastructure burden.

Before the cottage

Long before VeloCloud, Uppal studied electrical engineering at the Indian Institute of Technology Bombay. He earned a master’s in electrical and computer engineering at the University of Wisconsin-Madison and another in management of technology at Stanford. The combination suits the career that followed: a technical education with room for the question of how an invention becomes a business.

At Hewlett-Packard, he ran the Internet Infrastructure Division as vice president and general manager before leaving in 1999. He subsequently held engineering and business-development responsibilities at Webvan. Later came the president and chief executive role at Caymas Systems, followed by responsibility for application networking at Citrix after its acquisition of Caymas.

He also served as president and chief operating officer of OnMobile Global, the publicly traded mobile-services business. These jobs put him in several different parts of the technology economy: infrastructure, online commerce, enterprise access and mobile services. By the time he was a startup co-founder, he had already worked on both the technical product and the organization needed to sell it.

His life outside work had its own geography. In a 2011 interview, while at OnMobile, he discussed finding fellow cyclists and demanding terrain around Bangalore after years in the United States. He described cycling as a way to find his spirit. The detail gives the executive résumé some welcome texture: a person looking for a good ride, as well as the next market.

There is a teaching credit on his résumé, too: Uppal co-authored Software-Defined WAN For Dummies with Woo and Dan Pitt. A book with that title makes a particular promise to the reader. The subject may be complicated, but the reader should be allowed a way in. It is a useful companion to his later insistence on partner education, and another way of carrying an engineering idea beyond the people who built it.

A sales strategy disguised as homework

Uppal’s early commercial decision was to sell through partners. That meant other businesses would help bring VeloCloud to customers and operate the service. In September 2014, the company introduced a partner program with a portal for monitoring and managing customer networks. A seller used to supplying equipment could take on a continuing service relationship.

This matters because a network purchase rarely ends with delivery. Someone must configure it, watch it, answer questions and help when applications misbehave. VeloCloud’s partners could make the service available to companies that would never have encountered a small startup’s sales team. In return, partners needed something they could understand and support.

By May 2018, with VeloCloud inside VMware, Uppal was emphasizing training. Nearly 1,000 people had been trained since the acquisition, he told an interviewer. More informed conversations could lead to more trials and, eventually, customers. It was a notably practical account of growth: teach people how the thing works, then help them explain it.

“It’s all about training, training, training.”

Sanjay Uppal, 2018

The repetition is revealing. A new networking category needs people willing to trust it with everyday operations. The founder’s work therefore includes the patient business of making an unfamiliar idea ordinary enough to buy. An elegant architecture has limited commercial value if nobody feels comfortable installing it.

The exit that kept going

VMware announced its agreement to acquire VeloCloud in November 2017 and completed the deal on December 12. At the announcement, VeloCloud’s technology was deployed by more than 1,000 customers. VMware already had a substantial interest in software-defined networking inside data centers. VeloCloud extended that interest toward the branch office and the connections between locations.

The purchase also connected two different scales of organization. A startup’s product would sit within a vendor whose customers already depended on its infrastructure software. Explaining how those pieces fitted together became part of the job. The work after the announcement included sales integration alongside product integration, both subjects Uppal discussed publicly in 2018.

For Uppal, the transaction began another operating chapter. He became a VMware vice president and general manager, and later led its Service Provider and Edge business. His public discussions expanded to telecommunications infrastructure and edge computing, including work with Intel. The branch-network founder was now speaking about where computing would happen and how service providers could deliver it.

In 2018, CRN included him in its Top 100 Executives list, in the Disrupters category. He credited the VeloCloud team and its channel partners. The recognition arrived after the sale, a reminder that his public identity remained closely tied to the business he had helped start.

He could also be blunt. At London’s Great Telco Debate in December 2022, discussing Open RAN, he argued: “Open RAN will get there once we realise we have failed as an industry.” It was a pointed contribution in a room debating how much more open telecommunications infrastructure could become, and whether its economics would justify the effort.

The next traffic problem wears glasses

Broadcom completed its acquisition of VMware in 2023. Uppal subsequently led its Software-Defined Edge division and the VeloCloud business. In November 2024, he described another change in application behavior: AI systems were creating traffic patterns that conventional assumptions about branch networking did not comfortably accommodate.

His example was unusually tangible. At VMware Explore in Barcelona, he demonstrated Ray-Ban Meta smart glasses. A device sending video outward and receiving audio back reverses the familiar emphasis of watching streamed video. For a network designer, that changes the problem. The direction of traffic matters, as do bursts of activity and sensitivity to delay.

VeloRAIN, the architecture he outlined that month, proposed using AI to identify applications, prioritize their needs and improve network operations. The interest ran in two directions: using AI to manage networks and preparing networks to carry AI applications. Uppal was describing a product direction, with capabilities and development work, rather than announcing that every networking difficulty had been retired.

In April 2025, he discussed integrating fiber, cellular and satellite connections in a ZKast conversation with Zeus Kerravala. The available roads were multiplying. His recurring concern was how applications would travel across them without requiring the customer to manage every underlying difference.

VeloCloud abides

On July 1, 2025, Arista announced its acquisition of VeloCloud’s SD-WAN portfolio from Broadcom. Uppal joined Arista as a vice president and general manager. Arista’s switching and campus-network products gave the business a new setting, with branch connectivity fitting into a wider networking portfolio.

Reflecting on the move, Uppal remembered the early team in a cottage and a culture that valued irreverence and quick learning. His next ambition was to identify, secure and optimize traffic between AI agents across distributed businesses. The company had changed owners repeatedly; his attention kept returning to changing applications and the network adjustments they demanded.

He ended his reflection with a small piece of movie-inflected affection: “VeloCloud abides”. For a founder still accompanying his company through another transaction, the phrase carries some weight. There are grander ways to describe persistence. Uppal chose one with a wink. The connections still need tending, the applications keep moving, and somewhere a hotel IT manager’s perfectly reasonable question remains worth asking.