New Israeli Chip Firm Axonica Raises $50 Million, Aims to Challenge Nvidia

Axonica, a new semiconductor company established in June in Netanya's industrial zone, is attracting significant interest from venture capital investors. The company plans to develop an innovative chip for running artificial intelligence models and agents during the inference stage, aiming to be considerably cheaper and more power-efficient than Nvidia's graphics processors. Axonica also intends to compete with Cerebras Systems. The startup has already raised $50 million, with CEO and founder Assaf Tubul opening the round to further investors due to high demand, potentially reaching up to $80 million. The funding round is led by prominent US funds Amplify Partners, Wing VC, and Qualcomm Ventures, alongside Israeli funds Pitango, Viola, and 10D.

Tubul, a former senior executive at Qualcomm, was responsible for developing mobile communication chips and led AI innovation labs. He has extensive experience in low-power analog AI chips, model compression technologies, and AI model training algorithms, holding dozens of patents. Axonica is structured internationally as a US company with a development center in Netanya and a registered UK company in London. It currently employs about 20 people, ten of whom hold PhDs, an impressive workforce for an early-stage chip company.

The excitement surrounding Axonica comes amid fluctuations in the global and local chip market. Investors see significant promise in companies that can challenge Nvidia's dominance or those like Cerebras, which has faced stock value declines since its Nasdaq IPO. The sector has seen recent setbacks, including the dissolution of Paliops and the sale of Hailo to Microchip at a substantial loss. Consequently, many promising chip companies remain concentrated among specific investors like Avigdor Willenz and Gavin Baker.

Despite uncertainties in AI software business models, the hardware sector is benefiting from increased demand. Companies like Nvidia and AMD continue to grow, while their clients, including OpenAI and Anthropic, incur heavy losses, and cloud giants like Google and Amazon finance chip acquisitions and server farms with billions in debt. Many investors believe future competition will focus on power-efficient processors for AI inference, rather than expensive model training, potentially shifting significant computing activity away from Nvidia's processors.