Loading organizations...
§ Private Profile · Mountain View, CA, USA
Developer of networking chips and hardware for AI and machine learning workloads, enabling efficient I/O and memory movement with ACF technology.
Based in Mountain View, California, Enfabrica develops networking chips and hardware optimized for artificial intelligence and machine learning workloads in high-performance data centers. The company's Accelerated Compute Fabric technology, which includes the 3.2Tbps Millennium SuperNIC chip, addresses the scaling challenges of parallel infrastructure by enabling efficient I/O and memory movement across large-scale GPU clusters. Operating with a workforce of over 100 employees across North America, Europe, and India, the enterprise recently began shipping its Thames pilot systems to artificial intelligence infrastructure customers in early 2025. Enfabrica has raised $263 million in total venture capital funding, including an oversubscribed $115 million Series C round, drawing backing from prominent corporate and institutional investors such as Spark Capital, Nvidia, Cisco Investments, and Arm. The semiconductor organization was officially founded in 2020 by former Broadcom and Google executives Rochan Sankar and Shrijeet Mukherjee.
Enfabrica has raised $250.0M across 2 funding rounds.
Enfabrica has raised $250.0M in total across 2 funding rounds.
Enfabrica has raised $250.0M across 2 funding rounds. Most recently, it raised $120.0M Series C in November 2024.
| Date | Round | Lead Investors | Other Investors | Status |
|---|---|---|---|---|
| Nov 1, 2024 | $120M Series C | — | Alumni Ventures, Atreides Management, General Catalyst, Mangrove Capital Partners, Hans Tung, Partners Resolute, PICO Venture Partners, Seae Ventures, Brett Topche | Announced |
| Sep 1, 2023 | $130M Series B | Atreides Management | Alumni Ventures, General Catalyst, Mangrove Capital Partners, Hans Tung, Partners Resolute, PICO Venture Partners, Seae Ventures, Brett Topche | Announced |
Enfabrica is a silicon and software company developing high-performance networking solutions for AI infrastructure, specifically the Accelerated Compute Fabric SuperNIC (ACF-S) chip taped out in 5nm.[1][2] It serves AI system builders by solving I/O bottlenecks in scale-up and scale-out clusters of GPUs, CPUs, accelerators, memory, and storage, enabling elastic resource scaling, higher FLOPS utilization, and better resiliency compared to traditional fabrics.[1][2][3] The company targets hyperscalers and AI workload operators, with pilots underway in early 2025 and high-volume production following, backed by $115M funding raised in November 2024 to fuel expansion.[1][3]
Founded in 2020 (with some sources noting 2019), Enfabrica emerged from stealth to address exploding AI data demands, delivering multi-terabit-per-second, standards-based interconnects via 800GbE, PCIe Gen5, and CXL 2.0+ interfaces.[1][2][4] Its growth includes recognition in 2025 lists like TechRadarPro's hottest AI hardware companies and TechCrunch's top-funded US AI startups, culminating in a reported $900M acquihire by Nvidia in September 2025.[4][5]
Enfabrica was co-founded in 2020 by Rochan Sankar, ex-Broadcom executive who served as President and CEO, and Shrijeet Mukherjee, ex-SGI, Cisco, and Google with expertise in networking, as Chief Development Officer.[1][4] The idea stemmed from identifying critical scaling limits in accelerated computing infrastructures for AI, leading to a ground-up design of the "AI system NIC"—the ACF-S chip—to enable disaggregated, composable systems.[1][2]
Early traction built on this innovation: the company developed its modular, open-standards software stack alongside the hardware, achieving 5nm tape-out and positioning for customer pilots in 1H 2025.[1] A pivotal moment came with $115M funding in November 2024, accelerating market entry amid AI infrastructure booms, followed by Nvidia's acquihire in September 2025 to bolster its GPU networking capabilities.[3][4]
Enfabrica stands out in AI networking through purpose-built fabric technologies that outperform legacy solutions:
Enfabrica rides the AI infrastructure scaling wave, where exploding model sizes and data demands strain traditional networking, creating bottlenecks in GPU clusters and hyperscale data centers.[2][3] Its timing aligns perfectly with the post-2024 AI funding surge and 2025 hardware ramps, as providers like Nvidia seek composable fabrics for trillion-parameter models and agentic workloads.[1][4][5]
Market forces favoring Enfabrica include the shift to disaggregated compute (e.g., CXL memory pooling) and sustainability pressures for efficient FLOPS utilization amid energy constraints.[1][2] By solving these, it influences the ecosystem: its tech enables broader AI democratization, reduces TCO for cloud operators, and accelerates innovations like reasoning models—amplified by integration into Nvidia's stack post-acquihire.[4]
Post-Nvidia acquihire, Enfabrica's talent and IP will supercharge GPU networking for next-gen AI clusters, targeting exascale inference and multi-modal training.[4] Trends like memory disaggregation, agentic AI, and edge-to-cloud fabrics will shape its trajectory, with ACF expansions driving hyperscaler adoption.[1]
Expect deepened Nvidia ecosystem influence, evolving from standalone innovator to core enabler of elastic AI infrastructure—unlocking the scalability that defines Enfabrica's promise from day one.[2][5]
Enfabrica has raised $250.0M in total across 2 funding rounds.
Enfabrica's investors include Alumni Ventures, Atreides Management, General Catalyst, Mangrove Capital Partners, Hans Tung, Partners Resolute, PICO Venture Partners, Seae Ventures, Brett Topche.