The AI Inference process, which involves deploying educated AI models and apps, presents significant problems for enterprises. Using real-time AI data to solve tasks can be difficult and expensive, as AI accelerators have a track record of limited scalability and CPU bottlenecks can create system bottlenecks.
“Our disruptive AI Inference technology is unbound by conventional CPUs, GPUs, and NICs. We didn’t try to just improve an already flawed system. Instead, we unpacked and redefined the ideal AI Inference system from top to bottom and end to end, to deliver breakthrough performance, cost savings, and energy efficiency,” said NeuReality’s CEO Moshe Tanach, pointing to the paltry 30-40 percent utilization rate of AI accelerators.
Tanach said, “Investing in more and more DLAs, GPUs, LPUs, and TPUs…won’t address your core issue of system inefficiency.” It won’t get you to your destination any quicker than adding a faster engine to your car in order to get around traffic jams and dead ends. On the other side, NeuReality offers a huge AI pipeline an express lane, saving both capital and resources by smoothly channeling jobs to specially designed AI devices and quickly responding to your clients.
The European Commission’s venture arm, the EIC Fund, has demonstrated its support for the company’s solution, which improves energy efficiency in AI deployment while simultaneously delivering optimum performance, by investing in the EIC Accelerator program. Advanced semiconductors and artificial intelligence (AI), two sectors that are crucial to Europe and are predicted to be key contributors to economic growth in the years to come, are further addressed by the EIC investment.
The EIC Accelerator program awarded NeuReality a sizeable grant last year to help with a number of the firm’s innovative development stages that were intended to address the complexity and cost of AI Inference at the fundamental architecture level.
A seasoned executive team with substantial experience in AI, data-center architecture, systems, and software leads NeuReality, which raised its Seed round early in 2020. Together with Tzvika Shmueli, VP of Operations at Mellanox Technologies and former VP of Backend at Habana Labs, CEO Moshe Tanach, a former director of engineering at Marvell and Intel and AVP R&D at DesignArt-Networks (acquired by Qualcomm), and Yossi Kasus, VP VLSI at EZChip and former senior director of engineering at Mellanox, co-founded NeuReality. Lior Khermosh, the Chief Scientist and CTO of the company, was previously a fellow at PMC Sierra and a co-founder of ParallelM.