#StartupsEverywhere: Santa Clara, CA

#StartupsEverywhere: Vivek Raghunathan, CEO & Co-founder, Xscape Photonics

This profile is part of #StartupsEverywhere, an ongoing series highlighting startup leaders in ecosystems across the country. This interview has been edited for length, content, and clarity.

Data at the speed of light

Santa Clara-based startup Xscape Photonics is revolutionizing AI data center connectivity by developing multicolor silicon photonic lasers that drastically improve data center transmission efficiency. The company uses multicolor lasers to increase the speed of data transfer between chips and utilize the full capacity of existing chips, ultimately making data centers and other networked servers more efficient. We sat down with co-founder and CEO Vivek Raghunathan to discuss innovation coming out of the academic and research world, turning that innovation into a product for the market, how the government can support deep-tech startups, and his experience fundraising.

Tell us about your background. What led you to Xscape Photonics?

After earning my Ph.D. at MIT, I spent 15 years in silicon photonics, ranging from developing processor-memory packaging at Intel to leading commercialization at Rockley Photonics, and founding Broadcom’s Silicon Photonics division. In 2022, I joined Xscape Photonics as co-founder and CTO—later becoming CEO—alongside Columbia University professors Alexander Gaeta, Keren Bergman, and Michal Lipson, and laser pioneer Yoshi Okawachi.

We joined forces to translate their breakthrough multi-color laser research into a commercial platform, driven by a shared realization that traditional processor-memory bottlenecks waste energy and slow down performance. To date, we’ve raised $95 million.

What is the work you are doing at Xscape Photonics?

We're rewiring how chips in AI data centers communicate. Today’s advanced reasoning models and software agents require massive amounts of information to flow constantly between chips. When you ask AI a simple prompt, it processes data rapidly. But when you ask it to perform complex, multi-step reasoning tasks, processor efficiency drops by up to 100x. The processors sit idle waiting for slow data transfers, forcing operators to buy more hardware and build more data centers just to keep up.

Our solution is straightforward: high-speed, multi-color laser platforms that transmit multiple streams of data at the speed of light. By enabling instant data flow, our technology boosts processing efficiency by up to 100x while dramatically slashing energy consumption. We partner directly with accelerator providers to integrate our laser modules into their compute hardware before it reaches hyperscale operators.

Policymakers have talked a lot in recent years about the limits of the supply chain for semiconductors and related equipment, especially domestically and as tariffs complicate the ability to bring in foreign-made parts. How do you manage those risks at Xscape Photonics?

There has been more and more of a focus on ensuring that there is a diversification of the supply chain. Industry as a whole has been preparing for rising tariffs with an increased effort to bring manufacturing to the U.S. If you look at an optical component supply chain, there’s a high level of concentration in China. But there have been conscious efforts by the industry to enable new supply chains that avoid that reliance. There are new players that are emerging in other countries in Asia and also in the U.S., Mexico, and other neighboring countries.

But we’re starting to see that there’s a lag time when it comes to diversifying the supply chain with certain critical components. Many chip companies build with silicon wafers; others build on an indium phosphide wafer base. Shortages of these critical materials have impacted the ability for some to meet the demand for laser products that rely on those wafers; however, Xscape Photonics has designed our production to ensure that we are less reliant on a single supply chain partner–and less impacted by fluctuations in supply by leveraging volume manufacturing lines and a multi-vendor strategy.

What was the experience like taking innovation coming out of the academic and research world and turning it into a product for the market?

My co-founders all come from a research background, but they recognized the risk of moving directly from academia to industry, so they sought out someone like me with prior industry experience. I have the privilege of working with luminaries in the field who excel at the innovation and development the academic world demands and also understand the timeline and resources commercialization actually takes.

The end vision is to establish a data center where you have hundreds of lasers—each emitting 128 data streams via 128 different color lights—talking with one another and enabling communication between chips. Today the industry only uses one to four color lasers, and before customers place a bet on someone small like us, they want to see that our technology can actually scale in volume and work in the systems they already run. That’s why we came up with a completely new product—a much smaller laser that has four to 16 colors—as a go-to-market product. That will allow us to prove our technology and build trust with our customers without them having to make a billion-dollar investment to upgrade to a 128-color system.

What was your experience fundraising for such a technical and novel product?

The hardware world is challenging. This space doesn’t have software startups that go from launch to an enormous IPO in a few years. Manufacturing takes a long time, so betting on us means betting on a technology hardware manufacturing cycle that can put revenue five to seven years out.

We started off by focusing on customer relationships. Some of those potential customers, like NVIDIA and Cisco, then became investors and helped us bring in other investors, including venture capital firms. I learned that I had to seek investors who have the patience and conviction to invest in deep tech. The investors who have a true belief about the potential for deep tech innovations have the resources and advisors who can actually carry out technical diligence on the startups that they invest in.

Xscape Photonics recently announced that it is receiving funding from the Department of Energy through its Advanced Research Projects Agency-Energy SCALEUP Program. How can the government best support deep tech startups?

This is an exciting new program within the Department of Energy (DOE) focused on helping commercialize the innovation that came out of research the Department supported. We’re fortunate enough to get another round of support from the DOE, which helped fund some of my co-founders’ early research at Columbia. This kind of program is a huge win for startups like ours but also for the U.S. in general, especially as part of the ongoing efforts to bring back manufacturing to the U.S. and shore up the supply chain.

Outside of supporting the commercialization of research, the government should be more strategic about creating opportunities for startups. If you look at programs like the CHIPS and Science Act—arguably one of the biggest success stories of policy supporting the semiconductor ecosystem in the U.S.—the majority of that funding went to large companies with the firepower and administrative support to work closely with government agencies. You didn’t see a lot of startups like us, who lack the resources to dedicate a specific function to interfacing with the government. There should be more of a hybrid approach that allows startups to receive more of that funding. Large companies with the resources to navigate the federal funding process receive the award, and a portion of it flows to partnerships with startups. That would go a long way in helping startups bring innovative and problem-solving technology to the market much faster without having to necessarily work with the government directly.

Are there any local, state, or federal startup issues that you think should receive more attention from policymakers?

Right now, there is no unified federal framework guiding companies on AI deployment. Instead, we have fragmented and potentially conflicting state guidelines.

For a startup, navigating this lack of clarity means going well beyond what’s expected. To de-risk this regulatory ambiguity, we voluntarily adopted institutional-grade AI standards–investing in enterprise-tier AI infrastructure and achieving ISO certification early in our growth. However, ad hoc self-regulation isn’t a sustainable industry standard. A unified federal baseline is critical to give all businesses the clarity and certainty needed to deploy AI responsibly.

What are your goals for Xscape Photonics moving forward?

The long-term vision is to build and help our customers build the most efficient fabric of connectivity within AI data centers and help build the future of human intelligence hardware. To do that, we have to be attuned to what our customers need right now so that we—very similar to every other company—can get the revenue to help realize our bigger goals. Our focus is on getting our first product, a laser that we can plug into a server to make it work more efficiently, operating in our partners’ data centers in the next year and a half.

All of the information in this profile was accurate at the date and time of publication.

Engine works to ensure that policymakers look for insight from the startup ecosystem when they are considering programs and legislation that affect entrepreneurs. Together, our voice is louder and more effective. Many of our lawmakers do not have first-hand experience with the country's thriving startup ecosystem, so it’s our job to amplify that perspective. To nominate a person, company, or organization to be featured in our #StartupsEverywhere series, email advocacy@engine.is.