Lab-grade laser performance now fits on a single chip
For generations, the ultrafast laser has belonged to the laboratory — a powerful but unwieldy instrument too large and fragile for the wider world. Researchers at EPFL have now crossed a threshold long considered unreachable, embedding a high-energy pulsed laser onto a single photonic chip using an architecture called the Mamyshev oscillator. In doing so, they have not merely miniaturized a device but challenged the assumption that performance must be sacrificed in the pursuit of scale — a reminder that the boundaries of the possible are often drawn in pencil.
- Ultrafast lasers have been trapped in the lab for decades — powerful enough to drive atomic clocks and precision medicine, but far too bulky to go anywhere else.
- The core tension was physics itself: shrinking these systems onto a chip caused pulse energy to degrade, leaving integrated photonics stuck at a ceiling it couldn't break through.
- EPFL researchers cracked the problem by reimagining the Mamyshev oscillator — a fiber-based laser architecture — as something that could live in silicon waveguides etched onto a chip smaller than a postage stamp.
- The result is chip-level pulse energy that previously required a bench full of equipment, opening immediate pathways to portable atomic clocks, compact sensors, and next-generation telecommunications.
- The work, published in Nature, is early-stage but foundational — the field will now race to make this manufacturable, reliable, and scalable across industries hungry for ultrafast light at small scale.
For decades, the ultrafast laser — firing pulses measured in femtoseconds, a millionth of a billionth of a second — has lived exclusively in university labs the size of dining tables. Researchers at EPFL have now changed that, building a high-energy pulsed laser onto a single photonic chip and solving what many in the field considered a holy grail: preserving lab-grade laser performance at the scale of silicon.
The obstacle was always a collision of physics and engineering. Photonic circuits promised miniaturization, but pulse energy degraded as systems shrank. The energy that made ultrafast lasers valuable — for atomic clocks, precision cutting, medical procedures — seemed impossible to preserve once you moved from meters of optical fiber to waveguides etched into a chip.
The EPFL team's answer was the Mamyshev oscillator, a laser architecture that uses nonlinear optical effects to shape and compress light pulses. Proven in the lab, it had never been translated to integrated silicon — until now. By understanding how those nonlinear effects could be reproduced in chip-scale waveguides, the researchers produced pulses that would normally require a full bench of equipment.
The implications reach well beyond the lab. Atomic clocks — precise but large and expensive — could become portable and affordable, viable for satellites, telecom networks, and scientific instruments that currently can't justify the footprint. Integrated photonics underpins quantum computing, sensing, and next-generation telecommunications; any application requiring ultrafast light suddenly becomes feasible at smaller scale.
Perhaps most significantly, the work redraws an assumption the field had quietly accepted: that integration meant accepting lower performance. The EPFL results suggest the right architecture can preserve what makes ultrafast lasers powerful in the first place. The next challenge is manufacturability and scale — but the fundamental question of whether lab-grade laser performance could live on a chip has now been answered.
For decades, the ultrafast laser—the kind that fires pulses of light in femtoseconds, the kind that lives in university labs the size of a dining table—has resisted miniaturization. Researchers at EPFL have now cracked that problem. They've built a high-energy pulsed laser on a single photonic chip, using a design called a Mamyshev oscillator, and in doing so they've solved what many in the field call a holy grail: shrinking lab-grade laser performance down to something that fits on silicon.
The challenge has always been one of physics and engineering colliding. Ultrafast lasers produce extraordinarily short bursts of light—pulses measured in femtoseconds, each one a millionth of a billionth of a second. They're essential for precision work: cutting, welding, medical procedures, and critically, for atomic clocks. But the systems that generate these pulses have always been bulky. Getting that same performance into an integrated circuit, where everything happens on a chip smaller than a postage stamp, has eluded researchers. The energy in each pulse tends to degrade as you shrink the system. Photonic circuits, for all their promise, have hit a wall when it comes to pulse energy.
The Mamyshev oscillator is a particular laser architecture that works by using nonlinear effects in optical fiber to shape and compress light pulses. It's been proven in the lab. What the EPFL team did was translate that design onto a photonic chip—essentially reimagining how those nonlinear effects could happen not in a meter-long fiber, but in waveguides etched into silicon. The result is a device that produces high-energy pulses, the kind you'd normally need a bench full of equipment to generate, but now it's integrated into a single chip.
Why does this matter? Atomic clocks are the obvious application. Current atomic clocks are precise timekeeping instruments, but they're also large and expensive. A miniaturized version, powered by an integrated ultrafast laser, could make precision timekeeping portable and affordable. You could imagine atomic clocks in satellites, in telecommunications networks, in scientific instruments that currently can't justify the footprint or cost. But the implications spread wider. Integrated photonics is the foundation for next-generation telecommunications, for quantum computing, for sensing systems that need to be compact and efficient. Any application that needs ultrafast light pulses—and there are many—suddenly becomes possible at a smaller scale.
The breakthrough also represents a shift in how researchers think about photonic circuits. For years, the assumption was that you'd have to accept lower performance as the price of integration. The EPFL work suggests that's not necessarily true. By choosing the right architecture and understanding how to translate it to the chip level, you can preserve the performance characteristics that make ultrafast lasers valuable in the first place. The pulse energy limits that seemed fixed in integrated circuits have now been pushed higher.
This is still early-stage research, the kind published in Nature, the kind that labs will now build on. But it opens a door. Compact precision instruments—atomic clocks, yes, but also sensors, measurement tools, medical devices—suddenly become feasible. The telecommunications industry, which has long relied on integrated photonics for signal processing, now has access to ultrafast light sources at the chip level. The next phase will be taking this from the lab and making it manufacturable, making it reliable, making it something that can be produced at scale. But the fundamental problem—how to get lab-grade laser performance onto a chip—has been solved.
Notable Quotes
A holy grail of integrated photonics— Researchers describing the achievement
The Hearth Conversation Another angle on the story
What exactly is a Mamyshev oscillator, and why does it matter for this particular problem?
It's a laser design that uses nonlinear optical effects to shape and compress light pulses. In fiber-based systems, it's been proven to work well. The trick here was figuring out how to make those same nonlinear effects happen in the tiny waveguides of a photonic chip, where space is measured in micrometers.
So the challenge wasn't inventing something new—it was translating something that already works into a much smaller form factor?
Exactly. The physics was known. The engineering problem was: how do you get the same pulse energy and quality when you're working at a completely different scale? That's what took solving.
Why has this been so hard? Why haven't people done this before?
Because pulse energy tends to degrade as you shrink the system. The nonlinear effects that shape the light need space to work. Compress everything onto a chip and you lose performance. This design, it turns out, translates better than others.
And atomic clocks are the headline application, but what else becomes possible?
Anything that needs ultrafast light pulses and currently needs a lab-sized system. Precision sensors, certain medical devices, quantum computing platforms. Telecommunications networks could have ultrafast light sources integrated directly into their chips instead of relying on external lasers.
Is this ready to use, or is it still a proof of concept?
It's a proof of concept published in Nature. The physics works. Now comes the harder part—making it manufacturable, reliable, scalable. That's where the real engineering begins.