When Lisa Cummins Dulchinos joined Ayar Labs, she knew translating deep tech into a business story would be central to her role. At a company where 85% of the nearly 200 employees hold PhDs, Dulchinos found herself among experts fluent in micro-ring resonators and laser physics. To explain what Ayar Labs does, she took a different approach. “It’s basically a computer chip that allows us to open up the bandwidth in a computing system,” she tells us.
That chiplet—named TeraPHY—replaces traditional copper wiring with optics, using lasers to transfer data more efficiently. Dulchinos describes it with a relatable analogy: “There’s traffic on a highway…you open up and add more lanes.” This optical solution not only alleviates the data bottleneck but also delivers measurable benefits: “It decreases the amount of power required by four to eight times, increases the bandwidth almost 10 times, [and] reduces latency,” she tells us.
Read MoreWhat drew Dulchinos to Ayar Labs is also what defines her leadership: the intersection of technical complexity and strategic clarity. She frames the customer value in terms of “tokens per second, per dollar, per watt,” highlighting how improved throughput leads to greater profitability. Though the solution may cost more upfront, Dulchinos emphasizes its long-term value: “It decreases their total cost of ownership.”
By making the science digestible and the business case undeniable, Dulchinos reveals not just what Ayar Labs builds—but how finance can illuminate innovation.
Made Possible By
CFOTL: Tell us more about Ayar Labs. What type of company is it, and what is it about?
Dulchinos: Sure, —Ayar Labs is such an interesting company. It’s a startup—or rather, a 10-year-old startup—that’s solving the bottlenecks in AI compute architecture. What that means is, about a decade ago, our three founders—who came out of MIT and Berkeley—recognized that the volume of data flowing through compute systems was going to become a problem. They saw AI coming, and they realized the existing architecture, which relies on copper, wouldn’t be able to handle it.
Read MoreWe have a very technical team—around 85% of our nearly 200 employees are PhDs. If you ask them what we do, they’ll dive into micro-ring resonators and lasers. I try to explain it more simply: we make a computer chip, called the TeraPHY, that opens up bandwidth in compute systems, letting more data flow through. It’s powered by a laser, so we’re using optics instead of copper. Copper is power-hungry, and the power demands for future data centers are already unsustainable.
Our optical solution reduces power usage by four to eight times, increases bandwidth by nearly 10 times, and reduces latency. It’s kind of like a highway—if there’s too much traffic, you need to add more lanes. That’s what our chiplet does: it opens up more lanes so traffic (in this case, data) can move faster.
And from the customer’s perspective, that translates to a lower total cost of ownership. Even if our solution costs a bit more upfront, they can push more data—what we call “tokens”—through the system. That means more tokens per second, per dollar, per watt. Ultimately, it boosts profitability.
Ayar Labs | www.ayarlabs.com | San Jose, CA