LAS VEGAS — The term “AI factory” came up in multiple keynote presentations, media briefings, and in interviews with executives at this year’s Consumer Electronics Show (CES). (The show took place last week.)
And the concept seems to mean different things to different people.
“The term is absolutely being used in multiple — and often sloppy — ways right now,” said Alex Cordovil, Dell’Oro Group research director. In conversations with operators, hyperscalers, and infrastructure vendors, he said the main way the term is used is to mean “specialized data center.”
That’s also how Nvidia CEO Jensen Huang used the term. ”The reason I don’t call it a data center is that a data center is a place where people store data,” Huang said during a CES keynote.
“These AI factories are completely different from data centers,” Siemens CEO Roland Busch told reporters at the show. “They consume more energy, need liquid cooling, and cannot be controlled with normal building controllers — they need industrial-level controls.”
Booz Allen Hamilton CTO Bill Vass is on board with that particular definition. “An AI factory is a data center that is designed specifically for AI-specific hardware,” he explained. For example, it might have special concrete to support the physical weight of the server racks, and uses AI, digital twins and simulations to optimize power, cooling, and server placement.
Others, however, seem to use the word to mean something smaller than a data center, referring more to the servers, software, and other systems used to run AI.
For example, the AWS AI Factory is a combination of hardware and software that runs on-premises but is managed by AWS and comes with AWS services such as Bedrock, networking, storage and databases, and security.
At Lenovo, AI factories appear to be packaged servers designed to be used for AI. “We’re looking at the architecture being a fixed number of racks, all working together as one design,” said Scott Tease, vice president and general manager of AI and high-performance computing at Lenovo’s infrastructure solutions group.
That number of racks? Anything from a single rack to hundreds, he told Computerworld.
Each rack is a little bigger than a refrigerator, comes fully assembled, and is often fully preconfigured for the customer’s use case. “Once it arrives at the customer site, we’ll have service personnel connect power and networking,” Tease said.
For others, the AI factory concept is more about the software.
Some experts use the phrase in the sense of the AI technology stack. MIT Sloan Management Review’s Thomas H. Davenport and Randy Bean define it as a combination of “technology platforms, methods, data, and previously developed algorithms that make it fast and easy to build AI systems.”
“Not being able to build on an established foundation makes it both more expensive and more time-consuming to build AI at scale,” they wrote.
The European Commission also has a broad definition of the term; it envisions ecosystems that include computing hardware, human talent, and data with the end product being AI models and applications.
The EC also talks about “AI Gigafactories” — giant datacenters dedicated to training next-generation AI models. In 2025, the EU allocated €20 billion to create up to five AI gigafactories.
Finally, Deloitte defines an AI factory as a combination of hardware, software, and services that enable the AI lifecycle. “The primary product is intelligence, measured by token throughput,” said Nicholas Merizzi, a principal at Deloitte Consullting and the firm’s AI infrastructure US offering lead.
Just as a traditional factory takes in raw materials and generates a finished product, an AI factory takes in power, data, and large language models (LLMs) and outputs insights, he said.
Given how ubiquitous the term has suddenly become, it’s important to be clear on what’s being described when a vendor is selling an AI factory. Doe they mean the metaphorical kind? An actual building? A series of server racks? It’s the kind of misunderstanding that could cost a company billions if it invests in the wrong concept.
Editor’s note: Lenovo paid for Maria Korolov’s transportation and hotel costs for this year’s CES, but had no editorial role in the creation of this story.
Read the full article here
