The stakes of building a faster optical engine reach well beyond the inside of a data center. They reach into the power grid, the supply chain, and ultimately the climate. That is the argument Linkstar, a Singapore-based deep tech company, makes about its 1.6 terabit photonic engine. The device is small, but the problem it addresses is among the largest facing the technology industry.
Data centers have become some of the fastest-growing consumers of electricity on the planet. Within them, the optical links that connect servers and switches account for a meaningful and rising share of that energy use. As artificial intelligence drives demand for ever more bandwidth, the energy cost of simply moving data keeps climbing. The question is no longer only how to make computing faster, but how to keep the digital economy expanding without its energy appetite growing just as quickly.
Linkstar’s engine attacks that problem at its root. By cutting power consumption with each generation while carrying far more data per engine, the design moves more information for every joule of energy spent and for every gram of material used. The 1.6 terabit architecture markedly shrinks the module’s footprint and lowers its power draw compared with the previous generation. Those improvements are usually discussed as performance wins, but they are also environmental ones.
The benefits compound across the supply chain. A smaller footprint means less silicon, fewer discrete parts, and less electronic waste produced for every unit shipped. Lower power means less heat to remove, which lightens the cooling burden in facilities that already place heavy demands on local power grids and freshwater supplies. In regions where data centers compete with communities for electricity and water, reducing that burden is not a minor consideration. It shapes whether new capacity can be built at all.
For people, the payoff is the connectivity that modern life now quietly assumes. Responsive AI assistants, telemedicine, real-time collaboration, and the next generation networks that reach underserved communities all depend on the ability to move enormous volumes of data smoothly. Linkstar’s argument is that this connectivity can be delivered far more sustainably than today’s hardware allows, by making the underlying interconnect radically more efficient rather than simply adding more of it.
This is why the company frames efficient interconnect as one of the most leveraged yet least visible climate technologies in computing. Most discussions of sustainable technology focus on visible things like solar panels or electric vehicles. The plumbing of the internet rarely enters the conversation, even though it carries a growing share of the world’s energy demand. By building plumbing that is dramatically more efficient, the industry can let the digital economy keep growing without its emissions rising at the same pace. In effect, it can begin to separate the curve of human progress from the curve of power draw.
The timing gives the idea added weight. The current decade is defined at once by an artificial intelligence boom and a climate reckoning, two forces that can easily pull against each other. More AI tends to mean more compute, more data movement, and more energy. Technologies that break that link, by delivering the same capability for less power, are among the quietest but most consequential forms of environmental engineering available to the technology sector. They do not capture headlines the way a new model does, but they shape the long-term trajectory of how much energy the digital world consumes.
Linkstar’s chief executive, Dr Ben Yuan, describes the company’s purpose in those terms, building photonic engines that help AI and cloud data centers carry far more information with far less energy. Engineering light at scale, by this account, is not only a technical milestone but a contribution to a more sustainable digital future. It is offered as proof that focused deep tech ambition can move the needle for industry, for the planet, and for the billions of people who depend, often without realizing it, on the networks that a tiny engine quietly powers.
The broader principle behind the work is that the most enduring technologies tend to be the ones that remove a fundamental constraint rather than simply pushing against it. Efficient optical interconnect is becoming exactly that kind of foundational technology for the artificial intelligence era. It is easy to overlook, because it lives deep inside the infrastructure rather than on the screen. But as the demands of AI keep compounding, the ability to move data as light, faster, and with far less energy may prove to be one of the most important environmental advances the computing industry can make.
