NeuReality emerges from stealth to accelerate AI workloads at scale

NeuReality, a Caesarea, Israel-based startup developing high-performance AI hardware for cloud datacenters and edge nodes, today emerged from stealth with $8 million. The company, which counts among its board of directors Naveen Rao, former GM of Intel’s AI product group, says the funding will lay the groundwork for the launch of its first product later in 2021.

Machine learning deployments have historically been constrained by the size and speed of algorithms and the need for costly hardware. In fact, a report from MIT found that machine learning might be approaching computational limits. A separate Synced study estimated that the University of Washington’s Grover fake news detection model cost $25,000 to train in about two weeks. OpenAI reportedly racked up a whopping $12 million to train its GPT-3 language model, and Google spent an estimated $6,912 training BERT, a bidirectional transformer model that redefined the state of the art for 11 natural language processing tasks.

NeuReality aims to solve these scalability challenges with purpose-built computing platforms for recommender systems, classifiers, digital assistants, language-based applications, and computer vision. The company claims its products, which will be made available as a service, can enable customers to scale AI utilization while cutting costs, lowering energy consumption, and shrinking their infrastructure footprint. In fact, NeuReality claims it can deliver 30 times the system cost benefit over today’s state-of-the-art, CPU-centric servers.

“Our mission is to deliver AI users best in class system performance while significantly reducing cost and power,” CEO and cofounder Moshe Tanach told VentureBeat via email. “In order to make AI accessible to every organization, we must build affordable infrastructure that will allow innovators to deploy AI-based applications that cure diseases, improve public safety, and enhance education. NeuReality’s technology will support that growth while making the world smarter, cleaner, and safer for everyone. The cost of the AI infrastructure and AI-as-a-service will no longer be limiting factors.”

NeuReality was cofounded in 2019 by Tanach, Tzvika Shmueli, and Yossi Kasus. Tanach previously served as director of engineering at Marvell and Intel and AVP of R&D at DesignArt-Networks, which was acquired by Qualcomm in 2012. Shmueli is the former VP of backend at Mellanox Technologies and VP of engineering at Habana Labs. And Kasus held a senior director of engineering role at Mellanox and was head of very large-scale integrations at EZChip.

NeuReality has competition in OctoML, a startup that similarly purports to automate machine learning optimization with proprietary tools and processes. Other competitors include Deci and DeepCube, which describe their solutions as “software-based inference accelerators,” and Neural Magic, which redesigns AI algorithms to run more efficiently on off-the-shelf processors by leveraging the chips’ available memory. Yet another rival, DarwinAI, uses what it calls generative synthesis to ingest models and spit out highly optimized versions.

But Tanach says the company is currently active in three main lanes: (1) Public and private cloud datacenter companies, (2) solution providers that build datacenter solutions and large-scale software solutions for enterprises, financial institutions, and government organizations, and (3) OEMs and ODMs that build servers and edge node solutions.

“There are no such solutions in the market today. The competition is split between various silicon and system products. The most obvious ones are the inference deep-learning accelerators [from] companies such as Nvidia, Intel, and startups that are competing in that market. However, these competitors have only part of the solution both from a system perspective and from an AI compute capabilities standpoint,” Tanach said. “[We] will release more information about the solution later this year when its first platform is ready. For now, the company can only share that the total cost of ownership of its AI compute service will be more efficient by an order of magnitude compared to existing solutions.”

Cardumen Capital, OurCrowd, and Varana Capital led today’s seed round, the company’s first public investment. NeuReality has 18 employees.

By VentureBeat Source Link