Lenovo Unveils Massive AI-Ready Storage Portfolio Refresh

Lenovo announces new 'AI-ready' infrastructure systems in its largest-ever storage refresh - SiliconANGLE

Lenovo announces new 'AI-ready' infrastructure systems in its largest-ever storage refresh Lenovo Group Ltd. is targeting artificial intelligence with the largest-ever refresh of its data storage portfolio, announcing 21 new AI-optimized storage systems and infrastructure appliances for enterprise data centers. Announced today, the expanded portfolio fills the full spectrum of storage solutions, with new storage arrays, software-defined storage systems and virtualization offerings that have been crafted to tame the complexity of AI data operations and support real-time reasoning applications. The new lineup spans both Lenovo's ThinkSystem and ThinkAgile server models. The company promises a substantial performance boost to speed up AI inference, along with new features to simplify deployment and consolidate AI workloads, while reducing the total cost of ownership for AI applications, the company said. For instance, the new Lenovo ThinkAgile SDI V4 series encompasses a range of "full-stack, turnkey solutions" that aim to simplify AI infrastructure overheads and provide accelerated compute for data-driven reasoning applications. The company said testing shows its new servers can boost large language model inference performance by up to 2.4 times compared to its previous generation storage arrays. There's also a new generation of Lenovo ThinkSystems Storage Arrays, which provide a three-fold performance boost to accelerate actionable data-based insights for business operations. According to Lenovo, they also offer 97% greater energy savings and a 99% density improvement over its existing ThinkSystems arrays, ensuring a lower TCO for AI applications. Lenovo describes the new ThinkAgile and ThinkSystems servers as "AI-ready infrastructure" suitable both for enterprises that are just getting started with their AI implementations, and also for those with existing AI workloads. They're "purpose-built" to support any kind of task relating to AI, chiefly inference, but also fine-tuning and retrieval-augmented generation, as well as standard operations including databases, customer relationship management, enterprise resource planning, hybrid cloud, backup and archives. NAND Research Inc. analyst Steve McDowell said the performance boost in the new ThinkAgile and ThinkSystems arrays largely stem from the use of current-generation Xeon processors, bringing them in line with competing storage systems from rivals such as Dell Technologies Inc. and Hewlett Packard Enterprise Co. He also praised Lenovo for shrinking the overall density of the new servers. "Shrinking the footprint from the current generation's 8U rack footprint to 2U gives Lenovo a tremendous power and density story," McDowell said. "That's a 75% improvement in density, and it also delivers a huge improvement in power consumption. These days, every data center is feeling the power crunch, so this level of consolidation is a big win for Lenovo." Beyond these standard arrays, the company also unveiled a series of Lenovo ThinkAgile and ThinkSystem hybrid cloud virtualization systems, designed to provide flexible, independent scaling of compute and storage. These are aimed at enterprises that want to capitalize on AI while simultaneously optimizing virtualization. They include the all-new Lenovo ThinkAgile Converged Solution for VMware, which combines the enterprise-class features of the ThinkAgile VX Series with the robust data-management capabilities of the ThinkSystem DG Series storage arrays in a single package. For high-performance AI reasoning applications, Lenovo's new ThinkAgile HX Series GPT-in-a-Box solution may get some attention. It's said to be the industry's first liquid-cooled hyperconverged infrastructure appliance, built in partnership with Nutanix Inc. and designed for turnkey AI inference workloads, powered by Lenovo's proprietary Neptune liquid cooling technology. Capable of supporting distributed applications from the network edge to the cloud, it offers energy savings of up to 25% compared with earlier generation models. Lenovo adds that it supports 2.6 times faster transactional processing and 1.4 times faster input/output operations per second than earlier models, despite being squeezed into a much smaller form factor that provides 1.4 times better density. "The hardware isn't new, as the HX series is based on Lenovo's existing liquid-cooled ThinkSystems servers that were updated in September, but the bundling of its Neptune liquid-cooled servers with Nutanix's HCI and GPT-in-a-Box provides a nice option for power-constrained IT organizations when deploying turnkey AI applications," McDowell said. To help customers get up and running with its new servers and appliances, Lenovo is also offering new "AI starter kits," which provide a "validated, flexible and easy on-ramp" for AI inference and RAG workloads. All of the new systems come with integrated data security and resiliency features such as tamper-proof snapshots and software-based encryption to reduce operational risk and protect sensitive information. They also feature AI-powered autonomous ransomware protection, which uses machine learning models to detect and mitigate attacks. Lenovo's XCLarity systems management software is there to simplify set-up, management, optimize performance and streamline communication. McDowell thinks the refreshed storage systems will put Lenovo on a much more competitive footing versus its rivals, and it will likely become even more compelling once it closes on its pending acquisition of Infinidat Ltd. later in the year. "Lenovo has a strong entry-level and mid-range server story, and its higher-end systems will become much more powerful with Infinidat's technology," the analyst said. "I like the timing of this announcement, as we're going to see a slew of new storage offerings in the next 60 days at Dell Tech World, Pure Storage Accelerate and HPE Discover. So Lenovo is making a strong move by getting ahead of those rivals, and this should help to keep it relevant as those announcements happen." Scott Tease, vice president and general manager of Lenovo's Infrastructure Solutions Product Group, said the new server systems address questions around data sovereignty, compliance and availability, all of which are critical for AI applications. "Lenovo helps customers achieve faster time to value no matter where they are on their IT modernization journey with turnkey AI solutions that mitigate risk and simplify deployment," he said. "The new Lenovo Data Storage Solutions help businesses harness AI's transformative power with a data-driven strategy that ensures scalability, interoperability, and tangible business outcomes powered by trusted infrastructure."

Lenovo's 'Largest Storage Portfolio Refresh' Targets AI Readiness, Virtualization Costs

'A lot of customers in the SMB and midmarket, they don't have the ability to move off of VMware really. They're committed to that space, so helping them optimize is important,' Stuart McRae, Lenovo's executive director and general manager of data storage, tells CRN. Lenovo can save customers money on their virtualization deployments and get them ready for the AI era with the "largest storage portfolio refresh in company history," Stuart McRae, Lenovo's executive director and general manager of data storage, told CRN on Wednesday "The opportunity is for enterprises to deploy AI to accelerate their innovation and to show a solid ROI for that investment in AI. The challenge at the same time is to address the burgeoning cost of their virtualization environment. There have been lots of changes so that is really driving the focus on our innovation," McRae told CRN. For customers embarking on an AI path, he said they need high-performance storage to make all of their data actionable. That means storage that is designed to deliver faster performance, robust security, and resiliency to power the inferencing, retrieval-augmented generation (RAG), and agentic AI applications that organizations are hoping to deploy. "I think for many, most enterprises, making their data AI-ready is the essential first step in their AI journey," said McRae (pictured above). [RELATED: How Dell, Lenovo And Supermicro Are Adapting To Nvidia's Fast AI Chip Transitions] Among the new announcements are the Lenovo ThinkAgile SDI V4 Series, a full-stack turnkey device that simplifies IT infrastructure and accelerated computing that enhances time to value for LLM workloads. The company also announced the Lenovo ThinkSystem Storage arrays, three times faster than previous models, while using less power and offering a 99 percent density improvement when upgrading from legacy products. "That helps them in their licensing and their overall infrastructure costs," McRae told CRN. "Those products are designed to be up and running in minutes to make it easy for the customer and the business partners." McRae said in many cases Lenovo customers are replacing storage systems that are at least five years old, so those environments reap instant and massive gains from switching to modern, flash architecture, including a 97 percent reduction in energy costs. "Eighty percent of it is hard drive-based storage. If you want to modernize, to improve your virtualization, to be ready for AI you really need to move to flash and that's a big transformation for customers," he said. "When you consolidate you also have lower operating costs. You have fewer things to manage. Your service costs can be lower, because instead of deploying many boxes of storage, I've got it in this one case." McRae said it's particularly advantageous for small and mid-sized customers who can integrate all of their block, file and object storage onto one platform that comes with AI-driven ransomware protection. "They don't need multiple platforms," McRae said. "They can consolidate now all on one resilient flash solution." The ThinkAgile V4 HX Series GPT-in-a-Box features the industry's first liquid-cooled hyperconverged infrastructure (HCI) appliance that is capable of inferencing large language models (LLMs) with a 25 percent energy savings over previous generations. The full-stack generative AI product is part of Lenovo's partnership with Nutanix to deliver Nvidia NIMs and Hugging Face LLMs inside inferencing-capable hardware that can be deployed locally as a turnkey solution. "That creates a ready to run, pre-configured solution for running AI in a hyperconverged environment. You get all the ease of management and ease of use that HCI platforms are known for," he told CRN. "Some people talk about disaggregation. I think it's more helpful for customers to converge those solutions." For customers who are using VMware's bundled solutions, there are data usage costs that come with using vSan that customers would like to avoid. He said combining the HCI approach with appliance-based flash storage is a workaround that can save customers money. "Being able to offload your data to a high-capacity flash platform that's integrated into vCenter -- so you still get the same great usability and management of your data, but it can lower your licensing costs by 40 percent because we're moving that data on a super-efficient integrated flash solution -- I think that's very exciting," he said. "A lot of customers in the SMB and midmarket, they don't have the ability to move off of VMware really. They're committed to that space so helping them optimize is important." CRN has reached out to Broadcom for comment, but has not yet heard back. McRae said Lenovo is eager to see resellers and systems integrators take advantage of its ThinkAgile V4 HX series with Nutanix and ThinkAgile MX series with Microsoft's hypervisor to help customers save on virtualization costs. Additionally, as part of the storage announcement Lenovo unveiled AI starter kits -- packs of validated designs and reference architecture -- designed for its reseller partners to quickly deliver an AI systems to their customers. "We want our partners to do the consulting and to do the deployment and be that direct contact with the customer, but we want to help with 'where to start?'" he said. "We've designed these starter kits that are reference architecture, so the partner knows 'OK, well, I've got 16 GPUs. I need this level of storage. I need these Nvidia switches, this SR675 server with L40.' So, we give them that integration, the recipe, the list components, and how to put those together to simplify it for the partners."