UC Irvine engineers develop a revolutionary silicon wireless transmitter reaching 120 Gbps, matching fiber-optic speeds at ultra-low power.
What a Fiber-Speed Silicon Transmitter Actually Changes
Wireless links have long lagged behind fiber in raw throughput. Fiber carries huge volumes of data through glass with low loss over distance; radio links must squeeze bits through spectrum that is shared, noisy, and constrained by power and heat. A silicon wireless transmitter that reaches 120 Gbps closes that gap for short-range or high-capacity hops where running cable is costly, slow, or impossible. The point is not that copper and fiber disappear. It is that designers gain another option when they need fiber-class bandwidth without a physical fiber path.
Building that capability in silicon matters as much as the speed itself. Silicon processes are mature, manufacturable at scale, and already dominate digital systems. A transmitter that can be fabricated with familiar foundry flows is easier to integrate next to processors, switches, and radios than an exotic III-V or discrete RF module. Integration reduces board area, interconnect loss, and the number of components that must be qualified and cooled.
Why Ultra-Low Power Is Part of the Story
Throughput alone is a weak metric if the radio burns watts to deliver it. High-rate wireless links generate heat at the power amplifier, the mixers, and the digital baseband that prepares the waveform. Ultra-low power operation means the chip can sustain high rates without forcing large heat sinks, aggressive duty cycling, or power budgets that only fit in a rack-mounted chassis. That opens use cases where the radio sits on a board edge, in a small enclosure, or in a dense multi-radio system where thermal headroom is limited.
Low power also interacts with link design. When the transmitter is efficient, more of the link budget can go to modulation order, bandwidth, or antenna complexity instead of raw RF watts. Engineers can trade a fraction of that efficiency for longer range, better robustness under interference, or multi-stream operation. The reverse is also true: a wasteful high-speed radio often forces designers to shorten the hop or accept lower average rates.
Where Fiber-Class Wireless Fits in Real Systems
The practical sweet spot is high-capacity links that do not need the full distance of long-haul fiber. Think dense data-center aisles, building-to-building campus spans, temporary event networks, or backhaul for cells where trenching fiber is blocked by cost or timeline. In those settings, a 120 Gbps wireless hop can carry aggregated traffic that would otherwise require multiple slower radios or a cable pull.
- Replace or defer a short fiber run when install access is restricted.
- Add burst capacity between racks or rooms without rewiring.
- Provide failover paths that match the primary link’s order of magnitude.
- Prototype high-rate mesh or point-to-point designs before committing to cabling.
None of these replace careful RF planning. Line of sight, blockage, multipath, and regulatory spectrum limits still govern whether a fiber-speed wireless hop is usable. The chip solves generation and transmission of the signal at scale and efficiency; the environment still decides the link.
What Engineers Should Evaluate Before Betting on It
Treat the breakthrough as a component capability, not a finished product. Ask how the transmitter interfaces to baseband processing, what band and bandwidth it targets, and how receiver-side silicon, antennas, and packaging complete the link. Measure end-to-end latency, error rates under interference, and sustained throughput under thermal load—not only peak rate in ideal conditions. Confirm that the power figure includes the blocks you will actually ship, not a stripped lab configuration.
Also plan for the system around the radio: connectors, board materials at high frequency, shielding, calibration, and software that manages rate adaptation when the channel degrades. UC Irvine’s silicon wireless transmitter shows that fiber-class rates and low power can coexist in a mainstream process. Teams that want that performance in production still need to own the full chain from bits on the die to a stable link in the field.