Lenovo’s Hybrid AI Strategy for the Inference Era – Six Five Connected with Diana Blass

Models aren’t the bottleneck to AI adoption. Infrastructure is.

At CES 2026, Lenovo made clear that the future of enterprise AI will be defined by inference at scale, not experimental models. In this episode of Six Five Connected, host Diana Blass explores how Lenovo is tackling the real barriers to AI activation, from power and cooling to hybrid deployment across edge, on-prem, and cloud environments.

Featuring insights from Matthew Kimball, VP & Principal Analyst, Data Center at Moor Insights & Strategy; Vlad Rozanovich, SVP, ISG Sales at Lenovo; Flynn Maloy, CMO at Lenovo; and Ashley Gorakhpurwalla, EVP & President, ISG at Lenovo, along with perspective from Lisa Su, Chair & CEO of AMD; Yang Yuanqing, CEO of Lenovo; and Jensen Huang, CEO of NVIDIA, this episode breaks down what it takes to move from AI pilots to operational reality.

‍Key Takeaways:

🔹 Inference is the next AI wave: As Flynn Maloy notes, the inferencing wave is just beginning, shifting AI from model training to real-time operational decision systems.

🔹 Most enterprises are stuck in pilot mode: Matthew Kimball explains that outdated infrastructure and operational complexity are keeping organizations from scaling.

🔹 Hybrid AI is the enterprise default: Vlad Rozanovich outlines why AI must span edge, on-prem, and cloud to meet enterprise demands.

🔹 Infrastructure complexity is the real challenge: Ashley Gorakhpurwalla emphasizes that compute, storage, networking, security, and permissions must evolve together.

🔹 Liquid cooling is no longer optional: Kimball highlights Lenovo’s long-standing leadership in liquid cooling as chips grow denser and hotter.

🔹 Operational simplicity matters: Lenovo’s XClarity platform is designed to abstract complexity and help IT teams scale efficiently.

🔹 Hyperscale and enterprise are converging: Yang Yuanqing, Lisa Su, and Jensen Huang underscore how solutions like the AI Cloud Gigafactory with NVIDIA and Helios with AMD bring rack-scale AI to both hyperscalers and enterprises.

The winners in 2026 won’t be those with the flashiest models. They’ll be the ones who turn inference into real operations.

▶️ Watch the full video at sixfivemedia.com, and be sure to subscribe to our YouTube channel, so you never miss an episode.

‍Disclaimer: Six Five Connected is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

Diana Blass:
We came to CES to see what's really driving the year ahead.

Matt Kimball: 

My big, big takeaway is that AI is obviously front and center for everybody. 

Diana Blass:

But here's the catch. 

Matt Kimball: 

AI is adding so much complexity to the enterprise data center. And you've got these IT staff that are in place. Trying to manage ever complex clusters, infrastructure, and software stacks that frankly they don't have the resources or the skill sets to support.

Diana Blass: 

So what changes that?

Vlad Rozanovich: 

You can't just utilize your existing infrastructure that you have today.

Diana Blass: 

The answer isn't a better model, it's infrastructure.

Vlad Rozanovich: 

Compute, storage, network, security, permissions, it's all going to change.

Diana Blass: 

Outdated infrastructure has most companies stuck in pilot mode. They have proofs of concept, sandboxes, innovation labs. What they don't have is A.I. woven into operations. Think agents transforming supply chains, customer support or manufacturing, all working in unison across the business. Use cases promise under the next phase of A.I. known as inference.

Flynn Maloy: 

The AI inferencing wave is on its way. We're just beginning there. And when that comes, that's going to change how all businesses do it.

Diana Blass: 

Inference is when AI becomes alive inside a system, reacting, adapting, and making decisions in real time, continuously. But building that in the real world, where power is capped, cooling is constrained, space is limited, and reliability is non-negotiable, isn't easy. It requires a new kind of architecture.

Ashley Gorakhpurwalla: 

At Lenovo, we're solving for all these challenges with purpose-built infrastructure.

Diana Blass: 

At CES, that vision became concrete. From a new class of inference-built servers designed to bring compute closer to where data is generated.

Ashley Gorakhpurwalla: 

Internally, we call it the beast.

Diana Blass: 

To liquid cooling solutions that keep dense, always-on workloads from throttling under heat.

Ashley Gorakhpurwalla: 

You cut energy use by 40%, you boost performance, and you hit your sustainability goals.

Diana Blass: 

And architectures built for the enterprise reality, where privacy, compliance and control matter as much as performance.

Vlad Rozanovich: 

Now that we're starting to talk about enterprise, it is going to be this hybrid environment.

Diana Blass: 

Hybrid solutions place AI where it makes the most sense. on-prem, at the edge, and in the cloud. During Lenovo's keynote at CES 2026, we saw two new solutions that put hybrid into action. The AI Cloud Gigafactory with NVIDIA and the Helios architecture with AMD. The AI Cloud Gigafactory is built for hyperscalers operating at massive scale.

Yuanqing Yang: 

It helps the cloud providers to achieve time to first token much faster. and scale to hundreds of thousands of GPUs for trillion per meter large language models.

Diana Blass: 

Helios brings the same big AI power to enterprises. It's AMD's breakthrough, a super-efficient server rack that packs massive AI muscle into one cabinet, fast enough for huge tasks like analyzing company data or running smart assistants on the spot.

Lisa Su: 

With Lenovo adopting Helios, customers will have a clear path from on-prem enterprise workloads on the ThinkSystem SR675i to the most demanding training and distributed inference workloads on rack-scale AI systems.

Matt Kimball: 

If you look at what they've announced through their inference servers, through some of their storage announcements, and through their services organization is a holistic approach to take concept and turn it into actual activation with minimal amount of friction, with minimal amount of effort.

Diana Blass: 

Lenovo already operates in the world these platforms are built for, from edge devices to data centers to supercomputers.

Jensen Huang: 

That skill of manufacturing the systems, really complicated systems, and also installing these large systems into supercomputing centers is an expertise that Lenovo has.

Diana Blass: 

It also recognized early what now defined AI at scale, operational complexity and thermal limits. That's according to analyst Matt Kimball. Listen to this.

Matt Kimball: 

Liquid cooling is important. As you see, chips get hotter and hotter and more and more powerful. Liquid cooling is an absolute must. Lenovo has been leading the game in liquid cooling for at least the last 10 years. And I think there's a real distinction that Lenovo has established between itself and everybody else, quite frankly, when it comes to how it delivers this liquid cooling out to their servers. That's part one. Part two is XClarity. AI is adding so much complexity to the enterprise data center. XClarity is designed to abstract all of that complexity out and drive efficiency of operations through automation.

Diana Blass: 

So looking ahead, what we learned in this episode is this. The winners in 2026 won't be those with the flashiest models. They'll be the ones who turn inference into real operations. I'm Diana Blass, and now you're connected to how that happens.

‍