prnewswire.com web signal

DriveNets Raises $410M to Scale AI Ethernet Fabric

amd ai infrastructure chips ai-infrastructure

Key insights

  • DriveNets closed a $410M Series D on June 1, 2026, reaching $1 billion total capital raised while already cash-flow positive as of 2025.
  • AMD joined as a new investor alongside lead backers Bessemer Venture Partners and Atreides Management, with Red Dot Capital, Pitango, and D1 Capital Partners also participating.
  • The company reports over $1 billion in secured business from telecommunications operators and AI infrastructure providers before deploying this new capital.

Why this matters

The networking layer is emerging as a recognized bottleneck in AI infrastructure, and DriveNets securing over $1 billion in business before deploying this round suggests demand for high-performance Ethernet fabric has already outpaced available supply at scale. AMD's participation as a new investor validates multi-vendor open Ethernet approaches, a development that could shift procurement conversations at major AI operators running heterogeneous accelerator clusters rather than locking into a single vendor's interconnect stack. For founders and technical leaders, DriveNets achieving cash-flow positivity in 2025 signals that open networking infrastructure is a commercially viable model, not just an architectural counterproposal to vertically integrated offerings.

Summary

DriveNets closed a $410 million Series D on June 1, 2026, bringing total capital raised to $1 billion. The Israeli networking company entered the round already cash-flow positive as of 2025, with over $1 billion in secured business from telecommunications operators and AI infrastructure providers. The firm builds Ethernet-based fabric for large-scale AI deployments, targeting a problem CEO Ido Susan frames directly: 'The most expensive idle asset in the world right now is a GPU waiting on the network.' Essentially: (DriveNets, AMD, Bessemer Venture Partners) are treating open Ethernet fabric as the missing layer between raw compute and efficient AI operations at scale. - Bessemer Venture Partners and Atreides Management led the round; AMD and Red Dot Capital joined as new investors. - Existing backers Pitango and D1 Capital Partners also participated. - DriveNets supports multi-vendor AI accelerator environments, including AMD Instinct accelerators. Network throughput is increasingly the binding constraint in large AI cluster deployments, and DriveNets is now capitalized to scale inventory against that demand.

Potential risks and opportunities

Risks

  • If NVIDIA retains dominant share in large AI cluster deployments, DriveNets' multi-vendor Ethernet pitch may reach a smaller addressable market than the round size implies.
  • Scaling inventory ahead of confirmed orders introduces supply-side risk if AI infrastructure buildout timelines slip in late 2026 due to capital allocation shifts at major operators.
  • Telecommunications operators, a key named customer segment, have historically slow procurement cycles, which could extend DriveNets' path from secured business to recognized revenue.

Opportunities

  • Multi-vendor AI accelerator vendors, including AMD with its Instinct line, gain a credible fabric partner to compete against vertically integrated NVLink and InfiniBand deployments.
  • AI infrastructure operators running heterogeneous GPU clusters can use DriveNets to reduce idle GPU costs, an argument that strengthens as GPU hardware spend remains the dominant line item in AI buildouts.
  • White-box switch manufacturers and open networking hardware OEMs positioned for Ethernet-based AI fabric deployments could see increased demand pull through DriveNets customer buildouts.

What we don't know yet

  • The press release does not disclose a post-money valuation, so figures cited in secondary reporting cannot be confirmed from DriveNets' own announcement.
  • Whether the $1 billion in secured business represents contracted revenue, letters of intent, or another commitment metric is not specified in the release.
  • How DriveNets' Ethernet fabric interoperates with proprietary AI interconnects in mixed-vendor GPU clusters is not addressed.