← BACK TO PROFILES

Veerbhan Kheterpal

CEO and Co-founder·Quadric·Burlingame, California·

Agents are a new computer, and the cloud can't feed them

In this interview

In this interview, Veerbhan Kheterpal, CEO and co-founder of Quadric, explains how a failed attempt to build an autonomous vineyard robot revealed that no chip could run its brain, and how that became a nine-year effort to build a software-defined AI processor platform for cars, laptops, printers, and every other device outside the data center. He describes the patience it took to reach a Series C, why a platform that updates with software survives each new generation of models, and how his own company became an early large-scale adopter of agents. He also argues that agents are a new kind of computer the cloud cannot feed, that engineers will build six variants in the time they built one, and that founders survive by embracing the journey and staying intellectually honest about their core bet.

Veerbhan Kheterpal
Veerbhan Kheterpal, CEO and Co-founder at Quadric.

Veerbhan Kheterpal is CEO and co-founder of Quadric, which licenses programmable AI processor technology so chipmakers can run AI on laptops, cars, printers, and other devices outside the data center.

Veerbhan and his co-founders built one of Silicon Valley's early Bitcoin mining companies, exited in 2017, then tried to build an autonomous vineyard inspection robot and found no chip could power its brain. That gap became Quadric, founded in 2017 and carried by patient seed investors through years of platform research. The company closed a $46 million Series C in 2026, bringing total funding to $90 million.

“They say Nvidia is an overnight trillion dollar success. They've been in the making for almost 40 years.”

Veerbhan's core argument is that agents are a new kind of computer, and the cloud will fall far short of the tokens they need, so AI has to run on the device.

"If you're too worried about the destination, it's not going to work," Veerbhan explains. "You have to wake up and embrace the journey one day at a time."

The conversation

In this conversation: Jacqueline Samira (CEO, Howdy) and Veerbhan Kheterpal (CEO and Co-founder, Quadric).

Recorded for the CEO Studio interview series. Interview recorded on 07/30/26.

This interview was recorded by CEO Studio, a media brand owned by Howdy (howdy.com). Transcript lightly edited for clarity.

Veerbhan Kheterpal

I'm Veerbhan Kheterpal, CEO and co-founder of Quadric. We enable companies to build AI chips for performing AI functions outside the data center: on laptops, cars, printers, and really every device.

Jacqueline Samira

When you say outside the data center, what does that mean?

Veerbhan Kheterpal

Imagine when you're using ChatGPT today, you're accessing a supercomputer in the cloud. What we do is enable that level of capability to be built into your device, which means no subscription, no token limits, just unlimited AI built into the device itself.

Jacqueline Samira

So that's when we have AI in our cars or in refrigerators. Is that you?

Veerbhan Kheterpal

Yes, we enable that tech. We work with chipmakers across these verticals. Say someone is making a chip for a car. You're not paying attention while driving, a pedestrian steps into the way, and the car applies the brakes automatically. That's AI in real time on the car, paying attention when you're not. That functionality needs to come from somewhere in the chip in that car, and we work with chipmakers to enable it.

Jacqueline Samira

Incredible. So how did you get into this? Tell us a little bit about your journey.

Veerbhan Kheterpal

I come from a deep tech background, same as my co-founders. We did several deep tech companies before Quadric. About 15 years ago, we were one of the very early Bitcoin mining startups in the Valley, almost stealth, mining Bitcoin for our shareholders. When we exited that company around 2017, we saw the trend of AI very early, with lots of excitement and lots of work happening in the data center. We looked at the edge, outside the data center, and saw a big opportunity, almost white space. Too many companies were working on AI in the data center, and AI outside the data center needed a similarly scalable solution. That's what we've created.

Jacqueline Samira

How did you see that was happening?

Veerbhan Kheterpal

We actually built a robot before this company. In the transition between our previous exit and Quadric, we built an agriculture inspection robot. Napa and Sonoma have a lot of vineyards, and we built a robot that would walk the vineyards, digitize the farm, and help improve yields. In doing that, we discovered it was really hard, impossible, to get a robot at the right price for that market. That's how we came to the core problem: if you can't build a brain for a robot that works autonomously in a field on low power, you can't enable that robot. And that brain needs technology that we've created along the way. So the inspiration came from attempting to build a robot that would be very autonomous, like the robotics happening today.

Jacqueline Samira

So you're building this robot to walk the fields and you're stuck, because what you need doesn't exist. And then you realize this is a huge problem and there are so many devices that need this.

Veerbhan Kheterpal

That's exactly right.

Jacqueline Samira

How do you start? Do you pick an industry and say, we're going to focus on cars? What does the starting point look like?

Veerbhan Kheterpal

Typically that's how companies in our space work, but I admit we were naive enough to go after a very big problem and later realize how difficult it was. It took us a lot of time. In other words, we wouldn't have done it if we had known how hard it was going to be. When you pick a vertical, you know the exact problem that vertical needs and can go solve it. We were creating a platform from day one, which I would not recommend anyone do. You have this general view of what can exist in a few years once you're done, and the complexity is unknown and grows every day. We faced those challenges, but we've gone through verticals. Our early adopters were in automotive, then other markets such as laptops and printers, and some more exciting ones this year.

Jacqueline Samira

What year was the agriculture robot?

Veerbhan Kheterpal

2016, 2017. Some of the VCs in the Valley straight up told us, if you guys are right, we're going to need this kind of tech for everything, but it's a question of timing. And we now see that happening. We were, let's call it, a decade ahead of when it's really needed.

Jacqueline Samira

And when did you start Quadric?

Veerbhan Kheterpal

2017, between the previous company's exit and founding Quadric. Then we did a seed round with patient investors who have stayed with us to this day. Our seed investors are still investing, doubling down on the company, and were very patient in the early days of R&D. When you build a platform like this, it is so foundational. You're solving a lot of problems from scratch for a new domain, and that takes time and very deep tech talent.

Jacqueline Samira

There are spaces where you get crazy growth out of the gate and raise a bunch of money, and other spaces where time is exactly what you need because of how much you have to build. You've been on this journey for nine years. Do you believe you started too early and now is the time, or that it was the right time to develop everything you needed to be at this moment?

Veerbhan Kheterpal

I think it was the right time. If you look at the evolution of any significant platform company, Nvidia, AMD, Intel, there was almost a decade where they were figuring out the right platform. They say Nvidia is an overnight trillion dollar success. They've been in the making for almost 40 years. There was a lot of competition even then. Many, many companies were looking at what's called edge AI. It's a very difficult, fragmented space in which to build significant revenues, and we have been one of the lucky ones to get this far. But I will say we can now see the ocean starting to boil.

Jacqueline Samira

You say you're one of the lucky ones. What do you attribute that to?

Veerbhan Kheterpal

Persistence. We never gave up. We stuck to our plan and took two iterations of our technology, some with early partners. We found early partners who were also patient. That's the part that goes into the luck. And then just persistence on the key problems that need to get solved. Like they say, the only way you die is to run out of money. So you don't run out of money.

Jacqueline Samira

You recently raised another round. Was it your Series A, Series B?

Veerbhan Kheterpal

We're Series C.

Jacqueline Samira

Series C. Incredible. What does the future look like now? What's day two?

Veerbhan Kheterpal

We're in a phase where we have early users of our technology, a handful of companies deploying it in their chips. They love us and they're coming back for more, for their next generation chips, talking about their next set of applications and how our roadmap aligns. What we're seeing is that AI outside the data center used to be very specific applications: a car applying brakes automatically, image processing on a printer or a camera. It's getting to a new level, which we'll be talking about more in the next year, where you have agents, a very useful incarnation of AI. We've crossed the threshold where AI can reliably make a chained set of actions on your behalf: read your email, write an email, take actions, read a PowerPoint, do some work in Excel. These are extremely useful and incredible. That new application of AI is about six or seven months old. The revenues you see Anthropic and OpenAI scaling are agentic AI revenues that ramped in the last half year. Now we're starting to see that spill over outside the data center. Imagine the agent currently running in the cloud, which we access over our laptops, built into the laptop, or built into a box in a factory, making decisions and processing data. Literally a little brain, no cloud connectivity, independently doing work. That is turning out to be, I can safely say, a ten times larger opportunity than any of our other applications. The volume we plan to ship in this new application will dwarf everything we have ever done.

Jacqueline Samira

So six or seven months ago this became a possibility. Did you start developing for it then, or did you know it was coming and start years ago?

Veerbhan Kheterpal

Great question. Given the origin of the company, one of our product philosophies was to guard against the ever-changing AI domain. It's moving very fast. Models that came out six months ago are obsolete within the year. So we always built technology where we told customers, if you use us versus our competitor, we let you evolve the platform with software, as opposed to a hardwired solution that goes obsolete when AI changes. Which means when this new application came around, we were already ready. Our platform can adapt to it because it's a software-defined platform, not a hardwired solution for a specific kind of AI.

Jacqueline Samira

So you can change it with an update, unlike your competitors.

Veerbhan Kheterpal

Correct. That's one of the key features we've always had, and one big pillar of our success across these years. And when agents came around, our own experience using them was fairly disruptive inside the company. We're one of the frontier companies, even in the Valley, applying agents at very large scale. A lot of the work at our company today happens with agents. When you start doing that, you naturally ask, given that we're an AI infrastructure company, what do agents mean for us? These are all cloud agents, so the natural question is, what do on-prem or outside-the-cloud agents look like? That's been a very natural progression in the past six months.

Jacqueline Samira

How did you know you were one of the first companies deploying agents at scale? Was it talking to your board, talking to peers?

Veerbhan Kheterpal

I'd attribute it to the team. We've always played with all kinds of AI, given we're in the space, and a few of us noticed an inflection in capability around October or November last year. One thing led to another, with really crazy curious people at the company applying it to more and more use cases, until we heard, these guys are doing everything with agents. They're literally burning through tokens and producing stuff. It was natural discovery within the company, because we always encouraged experimentation: play with the latest AI, understand what's going on. We're in the space itself. I think the frontier labs have been planning this for a while. A lot of other participants are just getting aware. Given my conversations, even in the Valley, it's still very early. Very few companies are deploying agents at the scale I've seen.

Jacqueline Samira

If more and more companies do that, what does the future look like?

Veerbhan Kheterpal

This is me personally: the amount of infrastructure we're planning to build in the cloud, with all the capital expenditure and data center capacity, is going to fall far, far short of the number of tokens or the compute we'll need. These agents are not an application. They're like a computer. Think about how many computers we use in the world, between your phone, your laptop, the computer in the camera, computers in machines, embedded ones, big ones, small ones. An agent is a new computer. As more of these computers are replaced with agent computers, the cloud will be nowhere near enough with all the capacity being planned. In about twelve months, it's going to be obvious where we are and how big the token gap is. It will manifest as price increases in tokens, or we do something about it today.

Jacqueline Samira

What other manifestations will come from that?

Veerbhan Kheterpal

I call it a dislocation, and with that comes a ton of opportunity for various startups, consulting companies, infrastructure companies, OEMs, model companies. All have a role to play in how we reorganize the computing stack from what it looks like today to what it will look like with these agent computers. There's a lot shifting.

Jacqueline Samira

Let's shift gears a little bit. If agents are doing all of the coding, what do engineers do now? What is the future for engineers?

Veerbhan Kheterpal

I think it's more exciting. Let's assume enough tokens are available. There's a lot of talk today about companies budgeting tokens, and I think that needs to come off. Everyone needs to be empowered with enough tokens to unleash all their creativity. It's a very empowering notion: everyone with lots and lots of tokens, all engineers producing a lot more. You can throw away a lot of it, which allows for a lot of experimentation. An engineering team today can build one variant of a product in six months. They could build six variants in six months and use the best one. Much better products, created much faster. We've been through these cycles of super productivity jumps before, and humanity has found a way to employ more and more people every time. The rules will change and people will need to adapt. But moving from spending time crafting code to guiding the machine to craft the right code gets done much faster, so you produce a lot more production code and create more. We're going to see this abundance. We're going to get into an age of abundance where everything is available at much better prices and much more affordable.

Jacqueline Samira

I'm also an AI optimist. There was a period where I thought, is this bad? And I reflected on the history of humanity. Any time there's been a step change of massive innovation, it has disrupted the previous structure but created massive amounts more opportunity. Why would this era be different? Yes, it's going to massively disrupt everything, but it's going to create exponentially more, in an era where, if you can think it, you can create it. I completely agree that humanity is going to adapt. The path is going to be rocky, though.

Veerbhan Kheterpal

I agree, it always is. If you're going through a retraining period, certain types of roles will have to readapt, relearn, and get into other roles. But I think we'll work through that and find a better hill.

Jacqueline Samira

Every time we come to the Bay Area, it feels like San Francisco is living in the future. You go an hour or three hours outside the city, and the further away you get, the more people are in the present day. Say I'm a normal company and I want to be AI native, or AI first. What recommendations would you give a CEO looking at their org and saying, I want to be like that, but that's not my default setting?

Veerbhan Kheterpal

There's still a certain amount of barrier. I know exactly what you mean about Silicon Valley. There are barriers to adopting agents for personal use or company use. It's still not easy, and that's why you see frontier model companies partnering in a big way with consulting firms and PE firms, setting up a consulting layer and an army of people who can go in and enable everybody to adopt it. I've sat down with my board members and showed them what's possible personally, and they are shocked at the workflows that can be enabled. It's almost so magical that by themselves, people don't know what's possible or how to use it. We made this leap with AI, and everybody needs to catch up and start using it. It's no different from before. If you look at the internet in 1995, there were very few VCs, even here in the Valley, who understood what the internet was, and five years later there was a tech bubble. With agents, we're getting to that upswing, from early adoption and lots of use to massive adoption. I see it happening already.

Jacqueline Samira

We noticed something similar on our own team, where adoption was hit or miss based on a person's curiosity. We have eleven offices, and some were all engaged and some were far behind. So we hired a team, an AI researcher, an ML expert, a quantum computing expert, to build AI modules and courses and start hammering the team: this is what's possible, this is how we're doing it. People's natural curiosity comes out when they can observe it. One takeaway was that other companies, customers and non-customers, came to us asking, can you do these modules with our company too? And I thought, how are other companies doing this?

Veerbhan Kheterpal

We drive that proliferation of AI use at our company with lunch and learn sessions. Someone volunteers to present how they're using agents for their own workflows, and others pick up on it. But that still does not unlock the velocity required to empower all eight billion of us on the planet. One lunch and learn inside one company with 50 people doesn't do it. It's going to need a different level of catalyst.

Jacqueline Samira

I'd love some parting advice for CEOs who are in the thick of it. You've been working on this for nine years. It's sticking, it's growing, but you had to chew glass for many of those years. How did you get through it? What advice would you give CEOs in the first five years of their nine-year journey?

Veerbhan Kheterpal

The biggest thing is that if you're too worried about the destination, it's not going to work. You have to wake up and embrace the journey one day at a time, and say, a week from now I'm going to be compounding on this scale or that scale. It could be resilience, communication, fundraising, technical progress. But the focus on the journey is key to making it through. The second thing is to structurally always have a good anchor. I call it intellectual honesty. You have a certain idea about why your solution should work. At the very core of it, what are the benefits? They are unlocked because we do this, and we are the only ones doing this. So you have a certain argument for winning in the marketplace, and you are intellectually honest about it. That is not easy. Looking at your own solution and saying, these are the strengths, these are the weaknesses, these strengths are important and they fit into the worldview in this way. Then you have a structural reason to be there and to continue until things start to work. Businesses can take longer to build, sometimes slower. That structural spine is key, because you wake up and say, I have this core compounding principle on which I'm building this company. It's not hollow, it's not superficial. There's something very core here. That gives you the confidence to say, if I continue building, I am compounding something. If you define what that compounding is, then based on the exponent, it rises and starts going forward. It just depends on how long it takes. We've had tons of conversations in the last nine years where we removed noise from the table, this company doing this or that company doing that, by saying, this is our compounding principle, let's keep building that belief. The core bet is the important part. Then you work with that one day at a time, and that's what lets you keep going.

Jacqueline Samira

Very beautifully said. Embrace the journey, focus on what you can do in the moment that will compound, and be honest with yourself about where your company truly stands, the positives and negatives, where you can improve, and what your mission is. Thank you so much for your time.

Veerbhan Kheterpal

Glad to be here. Thanks for inviting me.

GET INVOLVED

Be part of the
conversation.

Whether you’re a CEO or founder who wants to be featured or a company interested in sponsoring, there’s a place for you here.