Israeli Chip Firm NeuReality Acquired for Up to $200 Million

Israeli chip company NeuReality, which develops solutions to reduce the operational costs of artificial intelligence systems, is being acquired by American firm Unconventional AI in a deal estimated between $150 million and $200 million. The transaction will involve a combination of cash and stock in the acquiring company, which has received investments from Amazon founder Jeff Bezos and several major Silicon Valley venture capital firms. While the official deal size was not disclosed, the financial range was reported by Globes. NeuReality, founded in 2019 by Moshe Tanach, Zvika Shimueli, and Yossi Ksus, had previously raised approximately $105 million. This sale is expected to provide investors with a significant return on their investment, though it falls short of the multi-hundred-million or billion-dollar exits seen in Israel's chip industry previously. According to PitchBook data cited by Globes, the company was valued at around $220 million in 2023, exceeding the current deal's valuation range.

Unconventional AI, the acquiring company, is much younger, having been established just last year. In December, it completed a substantial $475 million funding round at a $4.5 billion valuation, led by Andreessen Horowitz and Lightspeed Venture Partners, with participation from Sequoia, Lux Capital, and other funds, including a personal investment from Jeff Bezos. The acquisition has been approved by the boards of directors of both companies and is expected to be finalized in the fourth quarter of this year. The founders and development teams of NeuReality are anticipated to join the acquiring company, which intends to maintain its engineering operations in Israel.

The acquisition addresses a key challenge in the AI market: maximizing the utilization of expensive computing hardware. While companies invest heavily in data centers and advanced processors, the surrounding infrastructure for data transfer, inter-component communication, task distribution, and load management often creates bottlenecks. Processors can sit idle waiting for data or tasks, leading to increased costs for hardware, electricity, and cooling. NeuReality's approach focuses on optimizing this infrastructure rather than solely competing with AI accelerator manufacturers. The company developed the NR1 chip, based on 7nm technology, designed to offload system management and communication tasks from standard CPUs to dedicated hardware. They have since expanded into communication cards and management software. In March, NeuReality introduced NR-NEXUS, an operating system for managing inference processes in data centers, which is crucial as AI applications move beyond model training to widespread daily use by millions of users and businesses.