Lenovo Goes All-In On Hybrid AI—Even Software—At Lenovo Tech World

If a $57 billion company can fly under the radar in the tech world, Lenovo is that company. Obviously, everyone in the industry knows that it's a big and important player in PCs and servers (and in smartphones via Motorola), but I think that sometimes people pigeonhole it as "only" a reliable hardware maker, rather than giving it the level of attention of, off the top of my head, Meta, Microsoft, Apple, AMD, Intel, Google, Amazon—or, in this Age of AI, Nvidia.

However, the Lenovo Tech World event in metro Seattle last month showcased the most sophisticated version of Lenovo I've ever seen. Instead of just talking hardware and maybe discussing a few partners, chairman and CEO Yuanqing Yang and his executives put forward a comprehensive view of hybrid AI delivered at every layer from hyperscalers and the datacenter to the device at your fingertips. They talked about important partner initiatives with several of the industry titans I just mentioned (most significantly Nvidia). More than that, they put their work in the context of compelling use cases for different verticals and for consumers. And the biggest differentiator? The software that Lenovo is creating in-house.

The breadth of announcements at the event was a reminder of how many areas Lenovo operates in, and what it can really do. I won't detail every spec for every new product here, but rather dig into the big plays Lenovo is making and their strategic implications.

Making AI Pervasive—And Smarter—Across The Stack And Across Industries

Lenovo invoked its "Smarter AI for All" catchphrase time and again at the conference—but how is Lenovo delivering on that? First, the company is absolutely clear about its commitment to hybrid AI, which is music to my ears. I've been talking about a hybrid approach to computing for a decade. Now we see it in AI, with what Lenovo calls "a seamless, secure integration of private and public clouds, individual and enterprise solutions"—all the way to on-device AI tasks being performed on ThinkPads and Moto phones. Lenovo extends this to a "personal AI twin" to intuit your actions and improve your user experience.

It's worth remembering that Lenovo is one of a select few companies that can enable all the parts of hybrid AI. It has the computing stack, the industry partnerships and now the software to back it up. More details on that below, but at a high level I'll point out that Lenovo is wading into the enterprise SaaS area, aiming to deliver essential AI software through the cloud, locally on-device and everywhere in between. This was super unexpected for me—and I like it a lot.

The second thing to note about the delivery of "Smarter AI for All" is Lenovo's push into services. I've been talking about this for a while, but in short I'm impressed by what the Lenovo services teams are doing, especially as they build out specific use cases in the enterprise.

Recall that in the past few months, Wall Street has talked more and more about demonstrating the ROI of enterprise AI. Sure, everyone agrees that AI can do some impressive things; examples aren't hard to find. But do those cool examples stack up to a definable, calculable return on the high number of dollars invested in, let's say, cutting-edge Nvidia GPUs, or Amazon's Bedrock service, or watsonx from IBM? That's a much tougher question, and although it's still early days for enterprise AI, there's only so long that enterprise IT executives can keep saying "It's still early days" before CEOs and boards of directors reach for the ax.

I see Lenovo following a high-risk/high-reward approach to meeting this challenge using AI software and services. It's creating both horizontal and vertical plays that, if they work, could lead to big wins by giving strong answers to those ROI questions.

Building Out The Enterprise AI Stack, Leaning Into Open Source

Lenovo does a huge business in servers, and it has done a good job of penetrating the hyperscaler market, not coincidentally like its silicon partner AMD has. AMD CEO Lisa Su joined Yang on stage to talk up her brand-new server chips (the 5th Gen EPYC server processor and MI325X GPU, which I recently analyzed ) and to discuss the importance of AI generally.

"AI is truly the most important technology I've seen in my career," Su said. "We've made more progress in the last two years than we've made in the past 10." For Su—and I've had a number of conversations with her about this—an open ecosystem, including open AI models, and a spirit of collaboration are crucial for continuing to advance AI.

At the Lenovo event, her sentiments were echoed in a video recorded by Mark Zuckerberg. He talked about how the new Lenovo AI Now personal assistant, which I'll discuss below, is built on Meta's latest Llama model. Zuckerberg has become an advocate of open source ( something I covered in detail in September ), and in the video he said, "We [at Meta] believe that open source is the most cost-effective, customizable, trustworthy and performant option out there." Lenovo is clearly on board with that approach.

Yet Another Partnership For Nvidia, But With Crucial Differences

My subhead isn't meant to make light of the new Hybrid AI Advantage offering from Lenovo and Nvidia, which I think has some genuinely intriguing aspects. It's just that in the past few months I've sat in other conference halls for similar announcements when Nvidia CEO Jensen Huang joined Michael Dell to announce Dell Technologies' AI Factory offering , and then again when Huang joined HPE CEO Antonio Neri to announce their joint AI offering .

Mind you, all of this makes sense for both Nvidia and the partner OEMs. Nvidia commands 90%-plus of the datacenter GPU market and has industry-leading software (CUDA and NIM) to go along with it; Lenovo, Dell and HPE have massive footprints in servers, storage, PCs and so on. It's logical for the chip maker and the device makers to team up, at a high level because they can do more (and faster and better) together than they could separately, and at a nitty-gritty level because it allows enterprise customers to stick with their preferred datacenter vendors as they opt for full-stack AI solutions.

The real question will be which of these solutions differentiate themselves enough to break out of the pack. And I'm not talking only about the three I just mentioned. IBM, AWS, Google Cloud, Microsoft Azure and others would also be happy to talk to any enterprise about how their specific offering is the one to choose. It's important for Lenovo to be in the mix, but it will take time for this market to shake out. Some of the differentiation will come from factors beyond the raw technology deployed—things like use case definition and go-to-market actions.

What differentiates Lenovo Hybrid AI Advantage? The bottom levels of the stack were about like I thought: Lenovo ThinkSystem servers and ThinkStation workstations, Nvidia GPUs and networking, a joint data platform and the AI models and software (including NIM) you would expect. But on top of that is a layer of Lenovo services, including AI Fast Start and AI Advisory (consulting) to get enterprises ramped up. In line with what I just said about use cases and go-to-market, effective services engagements could make a big difference in enterprise AI deployments—and eventual ROI.

The top level of this stack may be the most interesting. It includes the new Lenovo AI Library, which is a collection of ready-to-deploy libraries and templates intended to speed up AI implementations, meshed with Nvidia NIM Agent blueprints. Different libraries cater to specific industries and functional areas. One example discussed at the event was legal review; while I haven't had a chance to vet these numbers for myself via Signal65, Lenovo claimed a 45% increase in accuracy and an 80% improvement in data reuse for the AI legal assistant it created. (Kudos to Lenovo for bringing out hard numbers here.) An AI marketing assistant was also discussed, and the company has cited other vertical examples including medicine and education, as well as business functions such as IT operations, product development and customer service.

The upshot of the Lenovo/Nvidia collaboration is that it lets any company onboard large language models and turn them into their own AI agents to perform specific tasks. As noted in our recent coverage of Salesforce's launch of its Agentforce platform , AI agents will play bigger and bigger roles going forward. Plus this represents one more way—a software-centric way, no less—that Lenovo is putting itself into the fray of enterprise AI.

There's not room in this piece to cover it in detail, but in connection with the stack diagram above I want to mention the sixth generation of the Neptune Water Cooling line that Lenovo announced at the event. It's no accident that all three of Lenovo, HPE and Dell have gone out of their way to talk about liquid cooling in the datacenter during the events I mentioned above. HPE even received a stock upgrade from Bank of America due to its prowess and patents in water cooling. The outsized compute requirements of AI mandate better cooling solutions. Lenovo touts the new version of Neptune as reducing power consumption in datacenters by up to 40%. Look for my colleague Matt Kimball to apply his expertise on liquid cooling to Lenovo's new offering soon. I'd also love to send the Signal65 Labs team to evaluate the tech.

The AI PC, Personalized AI And Lenovo AI Now

Lenovo's PC products have a legacy that reaches back to the dawn of PCs at IBM. Now Lenovo is working with partners including Microsoft, Intel, AMD and Qualcomm to make the transition from personal computing to AI-enabled personalized computing. At the event, it showed off a ThinkPad Aura laptop with AI-powered smart sharing, AI-assisted photo editing and lots of other engaging features. Rather than go into the product details here, I'll leave that to my colleague Anshel Sag (who has already reviewed a number of first-gen Copilot+ PCs, including the Lenovo Yoga Slim 7x ).

Although all of the Copilot+ PCs released so far have featured Qualcomm Snapdragon X processors built on the Arm architecture, that will change soon, and at the event Intel CEO Pat Gelsinger came on stage to talk about x86. Among other things, he discussed the new x86 Ecosystem Advisory Group that I covered in detail recently —of which Lenovo is a founding member.

Gelsinger, who led the design of the 486 chip back in the late 1980s, said that he believes having local AI on the PC will be more important historically than the PC receiving internet connectivity. He also cited a forecast that half the market will be AI PCs by the end of 2025, and all of it will be by 2030.

Yang said that the end goal of personalized AI that Lenovo is working toward is a personal AI twin—one assistant that functions across multiple apps and devices. Among other things, this will be supported by more immersive display capabilities that ease the use of 3-D models (including creating 3-D models from 2-D images), as well as wearables to help control the experience.

Lenovo exec Matt Zielinski took the audience through a couple of demos, both of which he tied to education, that really stood out for me. One was for Lenovo AI Now, the company's first AI personal assistant. This is a local AI agent that runs on your PC; it enables things like metasearch and RAG implementations at the device level, plus it has heterogeneous computing to process images and text at the same time.

He also brought a student on stage to help demonstrate Lenovo's education applications, which can turn an AI PC into a personal tutor. You feed the application relevant information, for example class presentations and notes, and it can spit out prompts for the student. At the event, Zielinski and the student did this by creating an AI-generated quiz based on the student's chemistry notes. Very cool—and immediately useful for lots of people.

AI For Good

From the start of the show, Lenovo also highlighted applications of AI to improve people's lives. Some of the most compelling uses are for people with disabilities. For example, Lenovo has developed an AI-powered solution for people with ALS that combines AI avatars and eye-tracking technology with LLMs. Voice synthesis can help ALS patients communicate with loved ones and caregivers even when they can no longer speak on their own.

The company also launched something called Alzheimer's Intelligence, an avatar companion for people with Alzheimer's and other forms of dementia. It uses curated real experiences to support these users. This product is connected to the company's Work for Humankind initiative, which I think is an admirable way to channel advanced technology into uses that really make a difference for people.

Lenovo's Breadth And Culture Match Its AI Ambitions

What we saw at Lenovo Tech World was a wall-to-wall program of AI from one of the heavyweights of the industry, and I walked away from it seriously impressed. I think it says something that this was the longest I've stayed at any event this year—most of four days—even though I have a company to run and lots of demands on my time.

Lenovo's culture is so strong and so good. At dinner or breakfast, the company's executives aren't hiding; they're sitting down with you and having real conversations. That extends to Yang himself, and I believe that culture spills into pretty much everything the company does.

On the technical side, I was most impressed by the in-house software Lenovo has created. That said, while the company is very strong in infrastructure for small and mid-sized businesses and PCs for the enterprise, it needs to be doing a lot more on infrastructure for the enterprise. To do that, it needs the product content, the proof points, the sales expertise and a lot more interaction with CIOs and enterprise operators. The upside is that from everything I know about Lenovo and all that I saw at Tech World, it really does have the goods—now it just needs to deliver.