Given the impact of photonics technology on modern life, it is still surprising how often I need to define the term when explaining what I do for a living. The response that seems to provide the most
immediate satisfaction in lay people is to tell them that photonics is the practical application of light. Why this should be is unclear. But photonics does seem more easily expressed by its applications than by its technologies.
Viewed strictly under the lens of technology, photonics can appear more like a branch of electronics. Emitters, detectors, and things like photovoltaic modules either require electrons to operate or rely on electrons to communicate their value, which is often parsed and measured in terms of things like wall-plug efficiency, sensitivity, or conversion efficiency.
Photons shine only in application. They are a source of both renewable energy and unique information and can alter materials with deft precision. Yet, for all that, photons have largely been viewed as a value-add in the age of electronics.
Given this historic role, it might be easy to miss the sea change that has started to blur the boundaries marking where electronics ends and photonics begins. This convergence is perhaps the inevitable conclusion of a progression that began when the platform for photonic components shrank from breadboards to circuit boards to silicon wafers and now discrete integrated circuit chips.
It is hard to pinpoint the exact moment when the photonic
integrated circuit (PIC) story became more about manufacturing
than research. Luxtera’s launch of the first commercial 10-Gbps optical transceiver in 2008 offers a likely candidate. But transceivers aside, integrated photonics is still in its early chapters. Whether the story applies to automotive lidar, biosensors, or even quantum computing, much of the action revolves around the challenge of scaling up production. Part of that challenge
is a cart-and-horse problem: PICs need sufficient market demand to justify investment in scaling production, but they need to scale production before the economies of scale make them an appealing new alternative.
With the acceleration of generative artificial intelligence (AI), PICs may have finally found their killer app. Data center electronics can train these models, but only so quickly and at a tremendous cost in power. Only photonics offers a sustainable solution to the chip-level interconnect required to mitigate these issues, and the manufacturing challenges surrounding PICs are largely what stand in the way.
It will become harder for electronics and photonics engineers
to overcome these challenges without learning each other’s
language. As their fluency grows, the comparative value of
electrons and photons inside a chip package will exceed the
sum of their parts. It will also become harder to distinguish
photonics and electronics at that scale, at least according to
traditional industry definitions.
Those definitions will not disappear. Applications for discrete industrial photonics solutions will continue. But a portion of
our industry, integrated photonics, is opening a new chapter.
Or perhaps, integrated photonics is the next chapter for the
semiconductor industry. Either way, as the historic barriers
between electronics and photonics engineering fade, expect
both to evolve into something rich and strange.
