IN Brief:
- Point2 Technology has expanded its Series B financing to $136 million, with Arm joining as a strategic participant.
- Its e-Tube platform transmits high-speed RF signals through plastic waveguides as an alternative to conventional copper and optical links.
- Funding will support Active RF Cable, near-packaged, and co-packaged versions of the architecture for accelerator interconnect.
Point2 Technology has expanded its Series B financing to $136 million as it moves its RF-based e-Tube interconnect platform from technology development towards commercial deployment in rack-scale AI computing systems.
The latest financing adds Arm as a strategic participant and was led by LB Investment, with continued backing from Maverick Silicon. Point2’s wider investor group includes semiconductor, connector, manufacturing, and infrastructure interests, reflecting the range of engineering relationships required to introduce a new physical interconnect into high-performance computing systems.
The company plans to use the funding to expand engineering, systems, operations, and commercial teams while accelerating three implementations of e-Tube: Active RF Cable, near-packaged e-Tube, and co-packaged e-Tube. Those formats move the RF interface progressively closer to the processor or accelerator and are intended to address different distances inside future AI systems.
The underlying technology transmits high-speed data as RF signals through plastic waveguides. Point2 positions the architecture between conventional high-speed copper interconnect and optical links, arguing that it can extend useful reach without the lasers and optical conversion required by photonic systems.
That middle ground is becoming more valuable as AI systems increase the number of accelerators connected within a rack. Higher lane rates make conventional electrical channels increasingly difficult to maintain over useful distances because insertion loss, reflections, crosstalk, and other signal-integrity effects grow with frequency. Equalisation, retimers, and active cable electronics can extend copper, but each adds cost, power consumption, or complexity.
Optical interconnect solves many of those reach problems but brings a different set of constraints. Lasers, optical engines, coupling structures, packaging, and thermal management all have to be integrated into a system that is expected to operate continuously and at very large scale. The engineering decision is therefore not simply whether optics can carry more bandwidth, but where the power, cost, serviceability, and manufacturing trade-offs justify introducing it.
Point2 is trying to create another option. An active RF cable can retain a pluggable cable architecture while using electronics to transmit the signal through a plastic waveguide. Near-packaged and co-packaged versions shorten the conventional electrical path further by placing the RF conversion closer to the compute package.
The company makes ambitious performance comparisons for e-Tube, including greater reach than copper and lower power, cost, and latency than optical interconnect. Those remain company claims rather than independent system-level benchmarks, so they are less useful than the architectural question they are intended to address: whether RF transmission through a low-cost dielectric medium can provide sufficient bandwidth density without inheriting the main weaknesses of either copper or optics.
Moving the interface closer to an accelerator creates additional engineering challenges. Near-package and co-packaged interconnect can improve electrical performance by reducing the length of the highest-speed copper traces, but they also increase pressure on package design, thermal management, manufacturing yield, connector strategy, and field replacement. A failed pluggable cable is straightforward to replace; a tightly integrated interconnect close to an expensive accelerator package is a different service problem.
Point2 is therefore developing multiple implementations rather than betting on one physical format. Rack architectures are still evolving quickly, and hyperscale systems are unlikely to settle immediately on a single balance between pluggable connectivity and co-packaged integration.
Arm’s involvement adds another connection to the processor ecosystem. Strategic investment does not amount to a design win, but it can give an interconnect supplier earlier exposure to the interface, packaging, power, and deployment requirements that shape future compute platforms.
The commercial challenge is substantial. New interconnect technologies have to meet demanding requirements for bit error rate, interoperability, manufacturing consistency, connector reliability, temperature behaviour, and long-term field operation. They must also be available at the volumes required by AI infrastructure customers and integrate cleanly with standards and system architectures that are still changing.
The $136 million Series B gives Point2 more resources to move through that qualification process. The next phase is no longer simply proving that RF signals can be carried through plastic waveguides. It is demonstrating that e-Tube can be packaged, manufactured, qualified, and deployed repeatedly enough to earn a place alongside the increasingly sophisticated copper and optical links already competing inside AI systems.
