By David Kuo, AVP, Product Marketing & Business Development, Point2 Technology
Semiconductor Engineering 2025 predictions
The demand for ‘all things AI’ is putting pressure on every link in the value chain – especially datacenters which need to address the soaring bandwidth requirements created by generative AI and ML while containing costs, expanding performance, and improving energy efficiency. It’s a tall order, but in 2025 we’re going to see a shift.
As the AI/ML workload evolves—pushing the limits on data rates with trillions of calculations processed every second—it has become vital that cabling interconnects keep pace to support the growth. This requires an evolutional shift from the copper and optical technologies we’ve historically used to innovative approaches like e-Tube, a revolutionary interconnect using plastic for terabit data transmission.
e-Tube as an interconnect technology uses mmWave RF signal to transmit and receive data over a plastic dielectric waveguide. e-Tube is capable of terabit level throughput without the expensive and power-hungry optical components and complicated DSP of optical cabling while exceeding copper cable length by 10X at similar cost structures. It is ideal for accelerator-to-accelerator compute fabric, top-of-rack (ToR) to Smart NIC, ToR to ToR, and Distributed Disaggregated Chassis (DDC) connectivity in the data center. e-Tube will scale to multi-terabit speeds and provides the bandwidth density to scale up AI clusters for many generations.
