Point2 Secures $76M to Advance RF Interconnects

Point2 Technology raised a Series B extension, bringing total Series B financing to $76 million, led by Maverick Silicon with participation from NVentures and UMC Capital. The company will use the capital to accelerate commercialization of its RF-based interconnect stack, centered on the Active RF Cable and its longer-term e-Tube family for near-package and co-packaged interconnect.

Point2 positions its platform around ARC, an Active RF Cable that uses RF signal techniques rather than conventional high-speed copper traces or optical transceivers. The company is explicitly targeting the trade-offs hyperscalers face with optical links, namely power, cost, and system complexity, while addressing the reach and latency limits of copper at high per-lane data rates. The announcement highlights two roadmap vectors:

• •ARC Active RF Cable for immediate rack- and tray-level reach and latency requirements
• •near-package e-Tube (NPE) for short-reach, high-density integration near accelerators
• •co-packaged e-Tube (CPE) for integration alongside compute packages in future scale-up modules

Point2 also emphasizes system-level engineering and integration work, not just link-layer PHY innovation. Expect substantial investment in signal integrity, thermal coupling, connector mechanics, and firmware for link management as part of productization.

Interconnect is re-emerging as a first-order constraint for large XPU clusters. As AI training and inference aggregate more accelerators into tightly coupled fabrics, latency, energy-per-bit, and density constraints force alternative approaches beyond copper and optics. Point2's RF approach is notable because it targets intermediate reach and low-latency use cases where optics can be overkill and copper is physically limited. The participation of NVentures and a lead from Maverick Silicon signals ecosystem validation from both accelerator and silicon-adjacent investors. Being named a BloombergNEF Pioneers winner reinforces strategic alignment with data center sustainability and scalability trends.

For system architects and hardware engineers, ARC and e-Tube represent a different point in the design trade space: they promise a blend of reach, bandwidth density, and energy efficiency that can simplify rack-scale topologies and reduce reliance on expensive optical switching in certain footprints. For ML infrastructure teams, this could lower rack-level power and cost while preserving latency-sensitive connectivity, affecting decisions on cluster topology, placement of parameter-sync domains, and co-location strategies.

Point2 will need to demonstrate production-grade reliability, connector ecosystems, and interoperability with accelerator vendors' PHY/serdes ecosystems. Key milestones to monitor are public performance numbers (BER, latency, reach at targeted data rates), thermal and mechanical qualification, reference designs with major hyperscalers or accelerator partners, and timing for NPE/CPE samples or demonstrations.

The funding round is a concrete signal that investors are betting on RF-based interconnect as a pragmatic complement to optics and copper for certain AI data center topologies. The work ahead is integration-heavy, but if Point2 can show low-power, low-latency, high-density links at scale, it will matter for future rack-scale and module-level system designs.