Point2 Technology, a San Jose company developing radio-frequency-based interconnects for accelerated compute systems, has closed a Series B extension led by LB Investment and backed by a slate of industry investors that includes Arm, NVIDIA and Bosch Ventures. The financing lifts the company’s total capital raised to $136 million.
Funding centered on rack-scale AI interconnects
The company said the new financing will be used to expand engineering, systems and operations teams to speed commercialization of products aimed at connecting accelerators inside AI racks and across rack-scale compute fabrics. Point2’s approach combines ultra-low-power mixed-signal system-on-chips with an RF signaling architecture intended to carry data over plastic waveguides, enabling what the firm calls Active RF Cables for high bandwidth density and reduced latency.
Point2 is led by CEO Sean Park. The firm describes its technology as targeting accelerator-to-accelerator links within data-center environments, where the performance and power characteristics of interconnects are increasingly critical for large-scale AI model training and inference.
Strategic backers reflect industry interest
The extension round was spearheaded by LB Investment and drew participation from a mixture of chipmakers, foundry-related investors and component suppliers. Named participants include Arm, Maverick Silicon, NVIDIA, UMC Capital, Molex and Bosch Ventures, among other new and returning investors.
- Lead investor: LB Investment
- Notable strategic participants: Arm, NVIDIA, Molex, Bosch Ventures, UMC Capital
- Total funding after extension: $136 million
The mix of backers spans IP vendors, accelerator makers and component suppliers, signaling cross-industry interest in alternative approaches to scaling data-center interconnects for AI workloads.
Product and market context
Point2’s platform centers on low-latency RF signaling across plastic waveguides rather than traditional copper or optical links. The company positions its Active RF Cables as delivering greater bandwidth density and lower latency for connections between accelerators within a rack or across racks — a problem set that has drawn significant attention as models and datasets have ballooned in size.
As hyperscalers and enterprise data centers pursue denser GPU and accelerator deployments, interconnect technology has become a pressurized area for innovation. Competing approaches include proprietary high-speed electrical fabrics, optical interconnects and co-packaged optics. Point2’s RF-based method represents an alternate route that emphasizes power efficiency and latency.
| Item | Detail |
|---|---|
| Lead investor | LB Investment |
| Strategic participants | Arm, NVIDIA, Molex, Bosch Ventures, Maverick Silicon, UMC Capital |
| Total after extension | $136 million |
The company did not disclose the exact dollar amount raised in the extension. Point2 says the proceeds will support product development and team growth as it works toward commercialization for future rack-scale compute systems.
Implications for AI infrastructure
Investments by major players such as Arm and NVIDIA suggest these firms are watching alternative interconnect technologies closely. Strategic funding can serve as both a validation of technical approach and a signal of potential ecosystem collaboration or integration down the line.
For data-center operators and cloud providers, the availability of cheaper, denser and lower-latency interconnects could influence rack design, accelerator placement and overall system power budgets. That, in turn, could affect how quickly providers can scale training clusters and deliver cost-effective inference services.
Point2’s path forward will depend on demonstrating the reliability, manufacturability and cost profile of its Active RF Cables at scale, as well as securing design wins with accelerator and server vendors.
Ravi Malhotra
Technology Editor, WE NEWS
