福联集成

Xsight Labs , a nine-year-old Tel Aviv startup that makes programmable data processing units and Ethernet switches for AI and cloud data center networks, raised a $300+ million round at a $2.8 billion post-money valuation. Fidelity was the deal lead, with Aliya Capital Partners, Atreides Management, Artisan Partners, Battery Ventures, Diagonal Capital, Intel Capital, Key1 Capital, Maverick Capital, Sienna, T. Rowe Price, Union Group , and Valor Equity Partners also participating. More here .

Founded in 2021, HuaChen Xinguang is an IDM-model developer and manufacturer of high-reliability semiconductor photonic (pump laser) chips used in optical signal amplification for backbone/core/metro networks, submarine fiber, inter-satellite laser links and AI data-center DCI. Mass production is slated for H2 2026.

Ayar Labs, founded in 2015 in San Jose, California, develops optical interconnect technology that uses light to transmit data between computer chips to address data transfer bottlenecks in AI systems.

CScale, founded in 2023 and headquartered in Palo Alto with ~85 staff and offices in Athens, Bangalore, Taipei and Shanghai, builds an integrated optical light engine for AI scale-up interconnect, replacing copper to link thousands of accelerators with high bandwidth and low latency while containing individual optical-link failures without interrupting compute.

Haiguang Xinzheng is a provider of silicon photonics chips and high-speed optoelectronic interconnect products, including optical modules, active optical cables (AOC), and high-speed active electrical cables. Its products are deployed in AI data centers to support high-speed, high-density data transmission. The company is among the few globally with both silicon photonics chip design capabilities and the ability to develop and mass-produce silicon photonic optical modules.

Delos Data is a Palo Alto-based AI infrastructure company co-founded in 2025 by former Barefoot Networks and Intel executive Ed Doe and Intel architect Dan Daly. It builds the Delos Nonstop AI product line, including a data interface chip, cluster reference architecture and servers, to remove network and data-movement bottlenecks in agentic AI inference.