Volantis, which aims to use vertical-cavity surface-emitting lasers, used in the iPhone's Face ID, to transmit data between AI and memory chips, raised $88M

The cost to integrate large AI models into existing hardware may decrease.

Volantis, a San Francisco-based semiconductor company, has secured $88 million in Series A funding. The round was co-led by Lachy Groom and Abstract Ventures, with significant participation from investors including John Doerr and Naveen Rao. The company plans to use these funds to advance the development and commercialization of its "A-1" photonic memory architecture. This innovative architecture combines photonics and semiconductor design to create a new category of AI inference systems. Volantis's technology utilizes a photonic interconnect and micro-VCSELs to link compute chips with memory, aiming to enable larger AI models, exceeding 20 trillion parameters, to operate at speeds of up to 10,000 tokens per second.

The substantial Series A funding for Volantis underscores a significant market interest in photonic interconnects for AI inference. This investment suggests a growing demand for hardware solutions that can overcome the memory bandwidth bottlenecks hindering the scalability of large AI models. The company's approach, using vertical-cavity surface-emitting lasers (VCSELs) similar to those in iPhone's Face ID, targets high-speed data transmission between AI processing units and memory, potentially offering a more efficient alternative to current electrical interconnects.

This development could accelerate the deployment of more complex AI models across various applications, impacting both cloud providers and end-users by reducing latency and increasing processing power. The success of Volantis's photonic memory architecture will be a key indicator for the broader adoption of photonics in mainstream computing and AI hardware, potentially setting a new standard for memory-intensive AI workloads.