Volantis Raises $88 Million to Tackle a Major AI Hardware Challenge
San Francisco-based semiconductor startup Volantis has raised $88 million in new venture funding as it works on technology designed to solve one of the biggest hardware challenges facing modern artificial intelligence systems: moving data quickly between AI processors and memory.
The company is developing a photonic technology that uses optical connections inside semiconductor systems instead of relying entirely on traditional electrical connections.
Why Memory Has Become a Major AI Bottleneck
Modern AI models require enormous amounts of data to be moved between computing processors and memory. Increasing the processing power of an AI chip alone does not necessarily solve the problem if the system cannot deliver data to the processor quickly enough.
Volantis argues that AI infrastructure is increasingly limited by memory bandwidth and the distance between computing chips and high-speed memory.
The company's approach is designed to shorten those limitations by using optical waveguides to move information between components.
Using Light Instead of Traditional Electrical Connections
Volantis is developing what it describes as photonic wires. These optical connections are designed to carry information using light through tiny waveguides integrated into semiconductor packaging.
The company says conventional electrical connections can travel only a few millimeters efficiently at the extremely high speeds required by advanced AI systems.
Its optical technology is designed to extend that reach significantly, allowing larger numbers of memory and compute components to operate together.
According to Volantis, its technology can support optical connections extending more than 200 millimeters on an interposer, compared with only a few millimeters for conventional electrical connections.
Designed for Extremely Large AI Models
Volantis is targeting AI systems that require huge memory pools and extremely high bandwidth.
The company says its technology is being designed to support models with more than 10 trillion parameters while maintaining low-latency access to memory.
Such systems could become increasingly important as AI companies develop larger models and more complex AI agents capable of performing long-running tasks.
Volantis says its architecture could combine very high memory bandwidth with significantly larger memory capacity than conventional approaches.
Technology Backed by AI and Semiconductor Leaders
The startup has attracted support from prominent figures in the technology and semiconductor industries.
Volantis says its backers include OpenAI CEO Sam Altman, Google Chief Scientist Jeff Dean, Meta Chief AI Officer Alex Wang and other technology and investment leaders.
The company's engineering team also includes people with previous experience working on high-bandwidth memory, advanced semiconductor packaging, photonics and optical communication technologies.
Why Photonics Is Becoming Important for AI
Optical communication has been used in data centers and telecommunications for years, but semiconductor companies are increasingly looking at ways to move optical connections closer to the processors themselves.
The reason is simple: AI accelerators are becoming faster, while traditional electrical connections face physical limitations when transferring massive amounts of data at high speed.
Photonic technology could potentially reduce some of these limitations by using light to move information between processors and memory.
Volantis is attempting to integrate those optical connections directly into semiconductor systems rather than depending on conventional external optical cables.
Potential Impact on AI Data Centers
If the technology reaches large-scale commercial deployment, it could have implications for AI data centers that require enormous amounts of computing and memory capacity.
AI companies currently spend heavily on GPUs, memory, networking equipment and electricity to support large-scale model training and inference.
Improving the way memory and processors communicate could potentially increase the amount of useful computation that can be obtained from expensive AI hardware.
Volantis says its technology is intended to improve bandwidth, memory capacity and energy efficiency for large AI workloads.
Volantis Focuses on the Memory Wall
The company describes its central challenge as the AI industry's growing “memory wall” — the gap between the increasing performance of processors and the ability of memory systems to supply them with data.
Traditional memory technologies such as high-bandwidth memory have continued to improve, but increasingly large AI models are creating new demands for capacity and bandwidth.
Volantis is attempting to address both issues through a photonic architecture that can connect a much larger number of memory components to AI processors.
$88 Million Funding Will Support Development
The newly raised $88 million will provide Volantis with additional capital as it develops and commercializes its technology.
The funding comes at a time when investors and technology companies are putting significant resources into AI semiconductor infrastructure.
AI computing demand has increased interest in technologies that can improve processor performance, memory access, networking and energy efficiency.
For smaller semiconductor startups, raising substantial funding can provide the resources required to move advanced hardware technology from laboratory development toward commercial products.
The Next Challenge Is Commercial Deployment
Although photonic technology offers potential advantages, turning advanced semiconductor concepts into mass-produced hardware remains a difficult engineering challenge.
Manufacturing yield, packaging, thermal management, reliability and compatibility with existing AI hardware will all be important factors as Volantis develops its systems.
The company's technology therefore remains an emerging approach rather than a replacement for the dominant AI accelerator infrastructure used today.
AI Hardware Race Expands Beyond GPUs
The Volantis funding highlights how the AI hardware race is expanding beyond faster processors.
Companies are now developing technologies across the entire AI infrastructure stack, including memory, networking, optical communication, advanced packaging and specialized accelerators.
As AI models continue to grow, the ability to efficiently move and store data could become just as important as raw processing power.
Volantis is betting that photonic connections can play an important role in solving that challenge.
What Comes Next for Volantis
With $88 million in new funding and a technology platform focused on high-bandwidth AI infrastructure, Volantis plans to continue developing its photonic systems for large-scale AI workloads.
The company's progress will depend on how effectively its optical architecture can be integrated into practical semiconductor systems and deployed at commercial scale.
If successful, photonic interconnect technology could become an important part of future AI infrastructure as models require increasingly large amounts of memory and faster communication between computing components.
For now, Volantis remains one of a growing group of semiconductor startups attempting to address the infrastructure challenges created by the rapid expansion of artificial intelligence.
Journalist: Vijay Singh