Next phase of NVIDIA Hybrid AI collaboration places transformational AI in the hands of every business

SAN JOSE, Calif. — (BUSINESS WIRE) — March 18, 2024 — Today, at NVIDIA GTC, a global AI conference, Lenovo announced new hybrid AI solutions, built in collaboration with NVIDIA, that deliver the power of tailored generative AI applications to every enterprise and cloud, bringing transformational capabilities to every industry. Stemming from the expanded engineering collaboration announced with NVIDIA during Lenovo Tech World, the innovations help enterprises and cloud providers with the critical accelerated computing capabilities needed to succeed in the AI era, taking AI from concept to reality and empowering businesses to efficiently develop and deploy new AI use cases that drive innovation, digitalization and growth.

The hybrid solutions are purpose built through engineering collaboration to efficiently bring AI to customer data, where and when users need it the most – from the pocket to the cloud – advancing Lenovo’s vision to enable AI for all and delivering time to market support of breakthrough architecture for the next generation of massive scale generative AI. Lenovo hybrid solutions, already optimized to run NVIDIA AI Enterprise software for secure, supported and stable production AI, will now also provide developers access to the just - announced NVIDIA microservices, including NVIDIA NIM and NeMo Retriever.

“Lenovo’s work with NVIDIA is pushing the boundaries of augmented intelligence for businesses everywhere, taking AI compute to where their data lives with a comprehensive portfolio of cutting-edge, hybrid AI solutions that will power generative AI from virtually anywhere while supporting sustainability efforts,” said Kirk Skaugen, President of Lenovo Infrastructure Solutions Group. “We are at an inflection point where new AI use cases are coming to market based on improvements in real-time computing, power efficiency and ease of deployment. Through our partnership with NVIDIA, we are delivering groundbreaking advancements in efficiency, performance, and cost that are accelerating AI applications across every industry and helping businesses immediately use insights from their data sets, whether for upgrading the retail experience, reimagining our cities, or unlocking the next level of smart manufacturing.”

“AI is a powerful force helping businesses unlock new insights from their data and enhance productivity,” said Bob Pette, Vice President of Enterprise Platforms at NVIDIA. “Lenovo’s new enterprise AI solutions, integrated with NVIDIA technology, demonstrate a pivotal milestone in supercharging compute performance for AI, delivering hybrid systems that businesses can rely on to power generative AI from virtually anywhere.”

As industries around the globe seek to use AI for analyzing vast bodies of data, power efficiency remains crucial to making the rollout of these compute-intensive workloads accessible for all businesses. Lenovo has been at the forefront of enabling efficient, high-power computing without compromise, pioneering Lenovo Neptune^TM liquid cooling technology and ranking #1 on the Green500 list for cutting-edge designs powered by NVIDIA GPUs that accelerate computing capabilities while keeping things cool even in high-heat, multi-GPU environments.

Lenovo ThinkSystem AI Servers: Catapulting Generative AI Inference and Efficiency with NVIDIA

Marking a game-changing milestone in efficiently supercharging AI-workloads at scale, Lenovo unveiled the expansion of the Lenovo ThinkSystem AI portfolio, featuring two new powerful 8-way NVIDIA GPU systems that are purpose-built to deliver massive computational capabilities with uncompromised power efficiency to accelerate AI implementation. Engineered for generative AI, natural language processing (NLP), and large language model (LLM) development, with time-to-market support for the NVIDIA HGX AI supercomputing platform, including NVIDIA H100 and H200 Tensor Core GPUs and the all-new NVIDIA Grace Blackwell GB200 Superchip, as well as the advanced, next-generation NVIDIA Quantum-X800 InfiniBand and Spectrum-X800 Ethernet networking platforms.

Lenovo ThinkSystem AI servers with NVIDIA B200 Tensor Core GPUs are set to power the new era of generative AI. The NVIDIA Blackwell architecture’s generative AI Engine, alongside the faster NVIDIA NVLink interconnect and enhanced security capabilities, propel the data center into a new era. With up to 25X more real-time inference to accelerate trillion-parameter language models, B200 GPUs are designed for the most demanding AI, data analytics and HPC workloads.

With an ultra-efficient power usage effectiveness (PUE) of 1.1., the new Lenovo ThinkSystem SR780a V3 is a 5U system that uses Lenovo Neptune™ liquid cooling. By using direct water cooling of CPU and GPUs and NVIDIA NV Switch technology, the system can sustain maximum performance without hitting any thermal limits. For more than a decade, Lenovo’s industry-leading Lenovo Neptune direct water-cooling solution, which recycles loops of warm water to cool data center systems, enabling customers to realize up to a 40% reduction in power consumption and a 3.5x improvement in thermal efficiencies compared to traditional air-cooled systems. As an industry metric used to determine the energy efficiency of a data center, PUE and power consumption are among the top-tracked sustainability methods, according to the “ Uptime Institute Annual Global Data Center Survey 2021.” Because liquid cooling provides a more energy-efficient alternative to air, the system can drive higher sustained performance while consuming less energy. It also allows the ThinkSystem SR780a to fit in a dense 5U package, helping to conserve valuable data center real estate.

Also accelerating the portfolio, the new Lenovo ThinkSystem SR680a V3 is an air-cooled, two-socket system built for maximum acceleration for complex AI with Intel processors and a choice of NVIDIA GPUs. The highly accelerated systems deliver massive computational capability and use industry-standard 19-inch server racks, allowing for dense hardware configurations that maximize efficiency without occupying excessive floor space or requiring shelving.

Lenovo also showcased a new Lenovo PG8A0N – the ultimate 1U node for AI and featuring open-loop liquid cooling for accelerators, supporting the NVIDIA GB200 Grace Blackwell Superchip. The new GB200 arrives to power a new era of computing. The superchip delivers 45X faster real-time large language model (LLM) inference, 40X lower TCO, and 40X less energy. In close collaboration with NVIDIA, Lenovo will deliver GB200 rack systems that supercharge AI training, data processing, engineering design and simulation.

Customers can run NVIDIA AI Enterprise, an end-to-end cloud-native software platform for the development and deployment of production-grade AI applications, on Lenovo’s portfolio of NVIDIA- Certified S ystems. For high-performance inference on popular AI models from NVIDIA and its ecosystem, customers can run the NVIDIA NIM inference microservices included in NVIDIA AI Enterprise on Lenovo enterprise infrastructure.

The portfolio includes Lenovo’s XClarity Systems Management, giving businesses a simplified and centralized resource management system, as well as Lenovo’s Intelligent Computing Orchestration (LiCO), a unified platform that streamlines use of clustered computing resources for AI model development and training and HPC workloads. The portfolio also supports the 4th and 5th generation of Intel Xeon Scalable Processors and provides thermal headroom for future higher-power GPUs.

Co-Designed NVIDIA MGX Solutions: Bringing Custom AI, NVIDIA Omniverse and HPC to CSPs and Enterprise

Lenovo is the leading provider of workstation-to-cloud support for designing, engineering and powering NVIDIA OVX systems and the NVIDIA O mniverse development platform. To help companies quickly build and deploy a wide range of purpose-built AI, HPC and Omniverse applications, Lenovo is partnering with NVIDIA to build accelerated models faster using NVIDIA MGX modular reference designs. Leveraging the designs, cloud service providers receive customized models faster with the delivery of accelerated computing for AI and Omniverse workloads economically and at scale. Including time-to-market support for the NVIDIA H200 GPU, the systems will enable scientists and researchers to tackle the world’s most challenging problems by accelerating complex AI and HPC applications running terabytes of data.