The future of computing has a very old-fashioned problem: moving things around. An AI accelerator can perform a large amount of arithmetic, but it is useful only if data can reach it, leave it and travel to the next processor without burning an unreasonable amount of power. The copper traces that served computing for decades are running into distance and energy limits. Light is being asked to take over more of the journey. For Adam Carter, this is less a sudden revolution than the latest turn in a career spent watching optical communications move closer to the center of the machine.
Carter is president and CEO of OpenLight, a California company built around a precise bit of materials engineering. Silicon is excellent for dense circuits and passive optical paths. Indium phosphide is excellent for making the active pieces that create, boost and modulate light. OpenLight combines them in a foundry process, putting lasers, amplifiers, modulators and detectors on the same silicon photonic chip.
The engineering is intricate. Carter's business pitch is deliberately plain: make that process available to other designers. Customers can license a process design kit, select validated components, draw a custom photonic application-specific integrated circuit and send it through a production route at Tower Semiconductor. Some customers design alone. Others hire OpenLight to help. In either case, the ambition is to turn a scarce craft into a platform.
A physicist moves down the supply chain
Carter began with the material itself. He studied applied physics at the University of Portsmouth, then completed a doctorate at the University of Wales, Cardiff, researching reactive ion etching of III-V semiconductor materials. These are the compound semiconductors used to make the active optical devices that now sit at the center of OpenLight's pitch.
His first industry job, in 1989, was as a process and device engineer at BT&D, a British Telecom and DuPont venture. From there, his career moved steadily away from the lab bench without leaving the physics behind. He crossed product management, Asia-Pacific sales, business development and strategic marketing at Hewlett-Packard, Agilent and Avago. The progression gave him a view of the whole optical chain: how a device is fabricated, how a module is assembled, how a customer qualifies it and how a supplier earns enough margin to build the next generation.
In 2007 he joined Cisco. A year later he took charge of its Transceiver Module Group, the unit responsible for the small optical engines that send data into and out of networking equipment. During his tenure, Cisco bought Lightwire, a silicon-photonics startup. Carter was involved in the acquisition and integration. His group went on to release Cisco's internally designed CPAK 100G transceiver family using a silicon-photonics optical engine.
Material to market
BT&D, HP, Agilent and Avago: engineering, product, sales and strategy across the optical stack.
The Cisco scale test
Runs the transceiver group, helps integrate Lightwire and brings an internal 100G silicon-photonics module to market.
Turnaround and exits
Oclaro returns to operating profit before its $1.85B sale; Foxconn Optical Interconnect is later divested to Broadcom.
The open platform
Joins OpenLight, raises $84M and pushes a foundry-validated photonics library into production programs.
A former colleague later described Carter's Cisco style as an unusual combination of tight accountability and broad empowerment. Proposals that held up could get resources quickly. Goals were refined with what the colleague called high granularity. That managerial rhythm fits transceivers, a business in which a fraction of a decibel, a point of yield or a small packaging delay can move an entire product line.
Learning the economics of light
Carter left Cisco in 2014 to become chief commercial officer of Oclaro. The components maker was losing roughly $35 million a quarter when he joined its senior team. Over 16 quarters, the business reached operating income above 10 percent per quarter. Lumentum bought Oclaro for $1.85 billion in December 2018.
The turnaround mattered beyond the tidy ending. Optical components punish vague strategy. Suppliers spend heavily before demand arrives, then face price pressure just as volume ramps. Product portfolios accumulate. Manufacturing footprints become hard to change. Carter's work sat at the commercial junction where technical possibility meets customer timing, supply contracts and the discipline to stop funding the wrong thing.
He moved next to Foxconn Optical Interconnect, where he restructured sales and marketing, developed a new go-to-market plan and helped shape a strategy that preceded the group's divestment to Broadcom in late 2019. Then he stepped away. The intended six-month break grew longer when the pandemic arrived. Consulting brought him back gradually. A venture investor he worked with knew OpenLight was looking for a CEO and put the opportunity in front of him.
Carter already knew OpenLight's technology and engineering team. The new question was whether a platform built inside Juniper could become infrastructure for everybody else.
The product is access
OpenLight came out of a long technical lineage. Its core work began at Aurrion, a UC Santa Barbara spinout founded in 2008. Juniper acquired Aurrion in 2016. Synopsys and Juniper formed OpenLight in 2022 to make the integrated process available to the market. When Carter joined as the company's first CEO in January 2023, he was walking into technology he understood and a commercialization problem he had seen from both sides.
At Cisco, owning silicon photonics could provide a proprietary advantage. At OpenLight, the advantage depends on sharing it. A customer pays for the right to use the PDK, can buy design help, pays production license fees and generates royalties as wafers ship. OpenLight does not need every customer to want the same chip. It needs many customers to trust the same process.
That distinction changes what counts as a product. A modulator model is a product. A tested laser building block is a product. So is a clean handoff to a foundry, a wafer-level test flow and a packaging partner who knows how to attach fiber without destroying the economics. OpenLight's partnerships with design houses, packaging firms and test suppliers are not decoration around the PDK. They are part of what makes the PDK believable.
The beachhead question
For a platform company, the uncomfortable customer question is obvious: who is already in production? Carter says OpenLight answered it plainly for years. The company was prototyping. In 2026, the answer changed. NewPhotonics, an early customer, placed volume production orders using OpenLight's process at Tower. It had designed its own photonic circuit with OpenLight's active and passive components and returned reliability data based on the same PDK.
Carter calls this a beachhead strategy. Land, establish evidence, then fan out. One customer's wafer order reduces uncertainty for the next buyer in a way a polished demonstration cannot. It proves that design files, process rules, devices, fabrication and reliability can make the whole trip together.
The company has not limited itself to customers that look alike. NewPhotonics used the library with relative independence. Oriole, a University College London spinout working on optical switching for AI workloads, brought system specifications and relied on OpenLight for chip design and tape-out management. The same platform can therefore serve expert teams that want autonomy and younger teams that need a guide.
The financing followed the production shift. OpenLight raised $34 million in August 2025 as it became independent from Synopsys and Juniper. In April 2026, it added a $50 million Series A-1 led by Matter Venture Partners. The announced use is practical: extend the library, build 1.6T and 3.2T reference designs, expand worldwide sales and technical support, and bring more customer programs to volume.
The quiet work after the breakthrough
Carter speaks about 400-gigabit modulators and integrated lasers, but he rarely leaves the manufacturing qualifiers behind. A component has to be testable. A package has to be assemblable. A process has to produce acceptable yield. A customer has to see a route from first tape-out to a supply chain that will still exist when demand rises.
That emphasis is the thread joining his career. The physics doctorate explains why he can discuss etching, optical gain and device tradeoffs. Cisco explains his appreciation for internal product integration. Oclaro explains the attention to portfolio economics and operating discipline. The break after 2019 gave him enough distance to recognize a different kind of opportunity: not one more optical component supplier, but a company that sells a reliable path through the component problem.
OpenLight still has to prove that an open photonics process can scale across customers and applications. The company is moving into production as AI data centers strain for bandwidth and as sensing and quantum systems ask for their own combinations of lasers, amplifiers and detectors. Carter's bet is that these markets do not need a single photonic answer. They need a common way to build many answers.
It is a platform thesis expressed in the language of a factory floor. Give designers choice, but validate the choices. Open the technology, but protect the process. Celebrate the speed, but count the watts, the yield and the fiber attaches. After 35 years in optics, Carter is trying to sell confidence that a beam of light can survive an organizational handoff.