San Francisco-based semiconductor startup Volantis has secured $88 million in venture capital to solve one of hardware’s biggest bottlenecks: connecting computing chips to memory.
📌 Core Breakthrough & Key Metrics:
- The Problem: Current GPUs from Nvidia and AMD are capped at around 8 High-Bandwidth Memory (HBM) chips per processor due to the physical range limits of electrical wiring.
- The Solution: Replacing copper wires with Vertical-Cavity Surface-Emitting Lasers (VCSELs)—the same proven optical technology that powers Apple’s Face ID.
- Massive Scaling: Using laser light enables packing 220 memory chips around a single GPU without reaching physical transmission constraints.
- Target Delivery: Volantis aims to launch its first commercial chip next year, significantly accelerating workloads like AI coding while avoiding exotic supply chain bottlenecks.
🤝 High-Profile Backing: Led by Lachy Groom and Abstract Ventures, the round attracted legendary investor John Doerr (early Google/Amazon backer), alongside VXI Capital, Susa Ventures, Naveen Rao, and Anthropic’s Sholto Douglas.
By marrying mature consumer tech with AI hardware, Volantis is positioning itself to revolutionize AI model memory scaling.