Lenovo details hybrid enterprise AI strategy, AI for PCs

Enterprise tech vendor and PC giant Lenovo is going all in on AI.

At the Lenovo TechWorld 2023 event today the company is detailing its vision, strategy, partnerships and some product direction for how generative AI and foundation models will permeate Lenovo’s portfolio. Lenovo has been incrementally adding AI capabilities to its lineup over the course of this year, including the launch of data management for AI in July and its TruScale edge service in September. 

At TechWorld, Lenovo is going much deeper outlining its strategy to build out hybrid foundation models that make use of public large language models (LLMs) fine-tuned on enterprise data with privacy controls, joining similar offerings from Cohere and Salesforce. Lenovo also has an aspiration of building out personal foundation models that serve the needs of individuals. 

As is the case with nearly every enterprise IT conference in 2023, there is also the obligatory partnership with Nvidia, including both hardware and software enablement for AI. Lenovo is also providing some early strategy insight into how it will be bringing the power of AI to PCs.

“We think the future of AI is the hybrid AI ecosystem,” Yong Rui, CTO, Lenovo group said during a briefing with press and analysts ahead of the TechWorld 2023 event.

Lenovo’s Hybrid AI strategy uses new techniques for training

Lenovo’s AI strategy is to benefit from the use of publicly available foundation models, as well as working on private and personal models.

In response to a question posed by VentureBeat about which LLMs Lenovo will be using, Rui said that his company will rely mostly on partners. Rui did not specifically identify the names of the public LLMs that Lenovo will use.

On the private AI side, Rui said that Lenovo is working on private foundation models that involve fine tuning and also enterprise knowledge with a vector database. Going a step further he emphasized that there is also a need to connect with existing enterprise systems such as enterprise resource planning (ERP) and customer relationship management (CRM) to really be able to complete IT enterprise tasks. To train the private models, Rui said that Lenovo has developed an approach he referred to as Joint Fine-Tuning.

Rui said Lenovo is working on technology to compress foundation models to make them more “lightweight,” so that they can run on personal devices, as opposed to the massive GPU clusters usually required. For training and enabling private models, Lenovo has developed a technique that Rui referred to as, joint model pruning and quantization for compression, that enables larger models to be optimized to work on personal devices.

A core component of the overall Lenovo Hybrid AI framework on the private and personal side is a data management and privacy proxy module. Rui explained that the module handles query understanding, encoding and decoding in an approach that enables and protects data privacy.

Nvidia and Lenovo partner for Hybrid AI

As part of the TechWorld 2023 event, Lenovo is announcing an extended partnership with Nvidia.

Scott Tease, VP and GM for HPC and AI at Lenovo commented during the press briefing that Lenovo has had a partnership with Nvidia for a long time in high performance computing. That partnership is now being extended to help advance Lenovo’s Hybrid AI vision.

“The goal of this initiative is to accelerate generative AI capabilities for enterprise clients everywhere,” Tease said.

Hardware is a key part of the extended partnership with the Lenovo ThinkSystem SR675 V3 server and ThinkStation PX workstation. The two systems integrate Nvidia GPUs and have been optimized to run Nvidia AI Enterprise software. Lenovo also announced that it has begun developing a new set of Nvidia Omniverse optimized hardware that will be available in 2024.

“The cloud is providing a massive amount of innovation to run AI, but the cloud is not the only place AI is going to exist,” Tease said.

Digital workspace and sustainability are part of Lenovo’s AI services strategy

As part of its AI efforts, Lenovo is also growing its professional services to help support AI deployment and adoption.

“We know that AI is really daunting for many companies, they’re figuring out how and where and why to use it,” David Rabin, Vice President, CMO of Lenovo Solutions and Services Group (SSG) commented. “But what’s most important right now is for them to identify opportunities, build the solution and implement at scale with speed.”

Lenovo is also using AI to bolster a new portfolio of digital workspace solutions branded as Care One designed to help improve workplace productivity. Rabin said that Care One provides highly automated and personalized support to users, powered by generative AI engines. 

Sustainability is another part of Lenovo’s AI-powered efforts as well. Rabin said that Lenovo has built a sustainability engine that empowers businesses to design their ideal IT environment. 

“It’s going to help you as an IT leader choose the best sustainability options,” Rabin said. “We then bring our consultants in with our AI technology and we help users visualize the influence of the business choices they are making and they impact their sustainability objectives.”

Lenovo sees a bright future for AI-enabled devices

AI is also coming to devices and a new generation of PCs, too.

During the press briefing, Daryl Cromer, Vice President and Chief Technology Officer of PCs and Smart Devices at Lenovo said that his company’s vision was to increasingly add AI capabilities to host of different devices, including PCs.

The AI powered PC, according to Cromer, will benefit from an AI assistant that helps users to do tasks on their devices, which sounds similar to the Windows Copilot announced by Microsoft earlier this month, though Lenovo has the advantage of controlling the underlying hardware, which Microsoft does not (in this case).

For Lenovo, Cromer sees AI helping to improve experiences and overall management of devices.  One such example is Lenovo device intelligence where an AI engine is being embedded to monitor the health of devices to see what’s going on, proactively identify issues and provide fixes.

“In summary, our goal is to put AI in your hands,” Cromer said.