Nvidia, Intel, and Amazon Eye Israeli AI Chip Startup DustPhotonics in Multi-Hundred-Million Dollar Deal

Israeli startup DustPhotonics is in advanced talks for a multi-hundred-million dollar sale. Discover why Intel, Nvidia, and Amazon are bidding for its 1.6 Tbps AI chip tech.

By: AXL Media

Published: Apr 13, 2026, 8:08 AM EDT

Source: Information for this report was sourced from CTech

Strategic Bidding War in Silicon Photonics

Israeli AI chipmaker DustPhotonics is at the center of a high-stakes acquisition battle, with industry leaders Intel, Nvidia, and Amazon identified as potential suitors. According to reports from Calcalist, the deal is expected to be worth hundreds of millions of dollars, reflecting the heating market for silicon photonics. This technology, which integrates optical components directly onto silicon chips, is becoming essential for maintaining high-speed data transfer while reducing energy consumption in massive AI data centers.

Willenz’s Track Record and Investor Confidence

The acquisition talks are bolstered by the pedigree of DustPhotonics Chairman Avigdor Willenz. Willenz has a formidable history of high-profile exits, having sold Habana Labs to Intel for $2 billion and Annapurna Labs to AWS. DustPhotonics has raised over $100 million to date from prominent investors including Greenfield Partners and Sienna Venture Capital. The startup’s Carmel series chips are currently at the cutting edge of the industry, supporting data transfer rates ranging from 400 Gbps to a staggering 1.6 Tbps.

The Pivot to Chip Development

Founded in 2017 by a team of veterans from Mellanox, Compass Networks, and Intel, DustPhotonics underwent a critical strategic shift in 2021. Under the leadership of CEO Ronnen Lovinger, the company moved away from manufacturing peripheral components like cables and transceivers to focus exclusively on core chip development. This transition positioned the company perfectly for the current AI boom, where optical connectivity is replacing traditional copper-based systems to handle the immense throughput required by generative AI.