Lenovo Group Ltd. is pushing to become the workhorse of the artificial intelligence industry after unveiling a slate of new, enterprise-grade server systems specifically for AI inference workloads.

The new servers are the latest additions to Lenovo’s Hybrid AI Advantage portfolio. They underscore how the AI industry is moving away from its focus on training increasingly powerful large language models to instead put them to use in real-world applications. Inference refers to the process where trained AI models make predictions and decisions and generate responses based on data and inputs they haven’t seen before —  where AI model rubber hits the road.

Lenovo said its goal is to help companies transform their significant investments in AI training into tangible business revenue. To do this, its servers are being offered alongside its new AI Inferencing software, which is focused on unifying and connecting organizations’ data from multiple environments so it can be leveraged by AI models.

The new servers are headed up by the Lenovo ThinkSystem SR675i, which is the company’s new AI inferencing powerhouse. It’s powered by Nvidia Corp.’s latest Blackwell graphics processing units and designed to run large language models anywhere at massive scale to support the largest AI workloads and accelerate AI simulation in industries ranging from healthcare to manufacturing and financial services.

There’s also the new Lenovo ThinkSystem SR650i, which provides accelerated AI inferencing power with high-density GPU compute and is designed to be installed in existing data center environments. Lastly, the Lenovo ThinkEdge SE455i is a compact model that’s aimed at deployments in smaller, edge locations in the retail, telecommunications and industrial sectors to enable AI inference at the edge. They’re built for low-latency and feature a rugged design that allows them to operate at temperatures between -5°C to 55°C.

Each of the new servers features Lenovo’s proprietary Neptune air- and liquid-cooling technology, and they’re available through the company’s TruScale pay-as-you-go pricing model, making them accessible to a broader range of enterprises.

Lenovo Infrastructure Solutions Group Executive Vice President Ashley Gorakhpurwalla said data needs to be transformed into AI insights the moment it’s created. “With Lenovo’s new inferencing-optimized infrastructure, we are giving customers that real-time advantage,” he said. “We are transforming massive amounts of data into instant, actionable intelligence that fuels stronger decisions, greater security, and faster innovation.”

Alongside the new servers, Lenovo announced new Hybrid AI Factory Services, including a new inferencing advisory that provides assistance in deploying and managing the hardware to optimize the performance of AI workloads.

Photo: Lenovo