San Francisco-based startup Volantis has secured $88 million in funding to develop advanced chip technology that overcomes the memory bandwidth bottleneck in AI processors by employing laser-based data transfer, promising faster and more scalable AI computation.
- Volantis raised $88M to innovate AI chip memory connections
- Laser-based data transfer could overcome current bandwidth limits
- Funding includes investments from notable tech and AI figures
What happened
Volantis, a semiconductor startup headquartered in San Francisco with connections to India, announced an $88 million funding round aimed at addressing a critical bottleneck in AI chip architectures. The investment round was led by Lachy Groom and Abstract Ventures, with participation from prominent investors including John Doerr and Angel investors from the AI industry.
The company plans to leverage vertical-cavity surface-emitting lasers (VCSELs), components widely used in the facial recognition capabilities of various consumer electronics such as iPhones, to create optical interconnects between AI computing chips and memory modules. This method aims to circumvent the electrical wiring limitations that currently restrict AI chip design and performance.
Why it matters
Modern AI chipmakers like Nvidia and AMD rely on surrounding their GPUs with expensive high-bandwidth memory chips to manage data throughput, but are limited by the physical reach of electrical interconnects. Nvidia’s best GPUs can manage only eight memory chips due to these constraints, which limits scaling and efficiency.
Volantis’s approach using laser light for data transmission could dramatically increase this capacity—potentially allowing up to 220 memory chips to be connected to a single GPU. This advancement could reduce costs, improve data speeds, and enable more powerful AI model processing, providing a significant competitive advantage in AI hardware.
What to watch next
Volantis aims to deliver a commercially viable chip product as early as next year that employs its optical interconnect technology to accelerate AI workloads, including coding assistance. Industry observers will closely watch the company’s ability to navigate advanced packaging challenges and supply chain issues, as well as the resulting performance gains in real-world AI applications.
Further funding rounds, partnerships with established semiconductor manufacturers or AI companies, and initial deployment of Volantis’s technology in mainstream AI hardware will be key indicators of the startup’s impact. The use of VCSEL technology in this novel application could inspire broader adoption of optical interconnects in future chip designs across the AI hardware sector.