|   | |
| Company type | Private company | 
|---|---|
| Industry | Telecommunications | 
| Founded | 2015 [1] | 
| Founders | 
  | 
| Headquarters | , | 
Area served  | Worldwide | 
Key people  | Ido Susan (CEO) Hillel Kobrinsky (CSO) [4]  | 
| Products | Network operating system, Software-defined networking | 
Number of employees  | 500 [5] | 
| Website |  www | 
DriveNets is a networking company that designs and sells infrastructure [6] to telecommunications providers, networks, and organizations building AI systems, including hyperscaler organizations and NeoClouds (GPU-as-a-Service providers).< [7] Products include a networking operating system running on a White box, and an Ethernet-based AI Fabric used for backend, compute, and storage networks.
DriveNets was founded in 2015 by Ido Susan and Hillel Kobrinsky. Susan is the co-founder of Intucell, which he sold to Cisco for $475 million in 2013. [8] Kobrinsky co-founded Interwise, which was acquired by AT&T for $121 million. [9] DriveNets was in a stealth mode [10] and was self-funded until 2019. [11] In 2019, DriveNets raised $110 million in series A round from Bessemer Venture Partners and Pitango Growth, [2] along with John W. Thompson and Stephen J. Luczo. [12] In 2021, DriveNets raised $208 million in series B funding led by D1 Capital Partners with follow-on investments from Bessemer and Pitango and investment by Harel Insurance. [13] In August 2022, DriveNets announced it completed Series C funding of $262 million led by D2 Investments, along with former investors Bessemer, D1 Capital, Pitango, Atreides Management, and Harel Insurance. [14] In 2025, AT&T acquired a 15% stake (approximately $750 million) in shares from DriveNets' employees and investors. The company's estimated value in 2025 was $5 billion. [15]
The company has approximately 100 customers, including AT&T [16] Comcast, [17] and KDDI. [18] Among its partners are Fujitsu, Broadcom Inc., [19] Itochu Techno-Solutions, Wipro, EPCglobal, [20] and Accton. [21] The company has 500 employees with offices in Israel [5] with offices in Israel, the USA, Japan, and Romania. [22] In November 2023, DriveNets joined the Ultra Ethernet Consortium, an industry initiative aimed at developing optimal Ethernet for high-performance networking. [23] [24]
DriveNets has two product lines - Network Cloud and Network Cloud-AI. Both are based on a Disaggregated, modular network architecture deployed on commodity white box hardware. The company’s Network Cloud product line is a cloud-native software-based solution that applies the architectural model of cloud shared infrastructure to communication service providers' networking, for a range of network use cases – from the core to peering to aggregation and to the edge, [25] provisioned as microservices and containers. [26] The company’s Ethernet-based GPU-to-GPU networking fabric product line is planned for hyperscalers and enterprises as a scalable replacement for InfiniBand in GPU cluster interconnections. [27]
DriveNets furnishes a network operating system (NOS) designed for cloud-native infrastructures [28] for building networks. [29] The Network Cloud architecture provides a software-based routing framework that allows networks to scale linearly and predictably. [30] The company has built software that runs on white boxes. [31] This network virtualization to a shared infrastructure [32] The company’s approach disaggregates the network, decoupling the control plane software from hardware. based on Open Compute Project DDC and Telecom Infra Project DDBR frameworks. [33] In 2023, it introduced an AI infrastructure networking platform, Network Cloud-AI, [34] based on Fabric Scheduled Ethernet (FSE). Fabric Scheduled Ethernet provides direct GPU connectivity in an architecture that uses standard Ethernet connections toward the client side but implements a hardware-based, cell-based scheduled fabric system to ensure predictable, lossless performance. [35] In November 2023, DriveNets joined the Ultra Ethernet Consortium to build an Ethernet Fabric for high-performance networking. [36] [37]