Enfabrica, a start-up based in Mountain View, California, closed a $125 million Series B financing round for its converged networking and memory fabric silicon and software.
Earlier this year, Enfabrica emerged from stealth mode to announce its new class of AI infrastructure interconnect chips called Accelerated Compute Fabric devices. Enfabrica’s ACF devices deliver unmatched scalability, performance and total cost of ownership for distributed AI, extended reality, high-performance computing and in-memory database infrastructure.
Enfabrica’s 8 Terabit/second ACF switching system, which directly bridges and internetworks GPUs, CPUs, and memories at scale with native, multi-port 800-Gigabit-Ethernet networking, is now accessible to customers for pre-order.
Enfabrica’s first, production-grade ACF switching system powered by Enfabrica silicon, produced in collaboration with partners, is an 8 Tbps platform enabling direct-attach of any combination of GPUs, CPUs, CXL-attached DDR5 memory and SSD storage to high-performance, multi-port 800-Gigabit-Ethernet networks. ACF switching systems incorporate Enfabrica’s high-performance ACF-S silicon having 100 percent standards-compliant interfaces and host networking software stack components running on standard Linux kernel and userspace interfaces. The ACF-S chip offers multi-port 800 Gigabit Ethernet networking and high-radix PCIe Gen5 and CXL 2.0+ interfaces.
The company envisions composable AI fabrics of compute, memory and network resources that scales from a single system to tens of thousands of nodes. The design provides uncontended access to >50X DRAM expansion over existing GPU networks via ComputeExpressLink (CXL) bridging.
“The fundamental challenge with today’s AI boom is the scaling of infrastructure,” said Rochan Sankar, Enfabrica CEO and co-founder. “There’s no denying the transformative value that AI delivers to a multitude of economic sectors. But there is a critical need to bridge the exploding demand to the overall cost, efficiency, and ease of scaling AI compute, across all customers seeking to take control of their distributed AI infrastructure and services. Much of the scaling problem lies in the I/O subsystems, memory movement and networking attached to GPU compute, where Enfabrica’s ACF solution shines.”
