Semiconductor giant Intel (INTC -9.20%) has hitched its wagon to the artificial intelligence (AI) revolution. The company is putting AI at the center of nearly everything it does.
In the PC market, Intel's latest Meteor Lake CPUs feature built-in hardware that enables AI inference workloads to be run locally. The task of blurring the background in a Zoom call, for example, can be offloaded to that AI hardware, which frees up the CPU and GPU to handle other processing tasks, providing a snappier experience and improving devices' battery life.
In the server market, Intel's Sapphire Rapids and Emerald Rapids server CPUs are also equipped with AI accelerators. While these CPUs don't handle AI workloads nearly as capably as high-powered GPUs from Nvidia (NVDA 6.18%), they offer an efficient way to run some AI inference tasks. Given how scarce and pricey data center GPUs have become, Intel's focus on lowering the total cost of ownership should appeal to its customers.
Intel also competes directly with Nvidia with its Gaudi line of AI accelerators. Gaudi2 is the most recent version, and while the chip isn't as powerful as Nvidia's top-tier product, it still delivers solid performance. The next-generation Gaudi3 will launch this year, providing a more compelling alternative to Nvidia's chips.
In addition, Intel can tap into demand for AI computing capacity with its foundry business. Intel is rapidly rolling out new process nodes as part of its roadmap. Intel 4 is in volume production now, Intel 3 will come early this year, and both Intel 20A and Intel 18A are expected to be ready by the end of 2024. With Intel 18A, the company expects to regain its manufacturing edge over TSMC. If all goes as planned, Intel will be a viable manufacturing alternative for AI chip designers.
Tackling AI software
Intel announced on Jan. 3 that it had teamed up with DigitalBridge to launch a new independent AI company called Articul8 that will provide a generative AI platform for enterprise customers. The technology behind the company's platform was developed at Intel, and both companies will maintain a close relationship as they go after the enterprise AI market.
Articul8's platform was launched and is optimized to run on Intel's server CPUs and its Gaudi accelerators, although the plan is to support a wide range of alternatives. Enterprises often use multiple cloud platforms and may use a mix of hardware, so supporting hardware from competitors will be critical for adoption. Still, as Articul8 gains momentum, Intel could see a boost in demand for its AI-enabled chips.
Articul8's platform looks similar to watsonx from IBM. Both companies have recognized that large enterprises have very specific needs when it comes to deploying generative AI technology. Security is a watchword for both platforms. Particularly in heavily regulated industries like financial services, keeping customer data private and proprietary data safe is paramount.
Another spin-off
Intel has an equity stake in Articul8, as do a cadre of venture investors; DigitalBridge serves as the lead investor. This isn't the first time Intel has sought outside capital for one of its business units. The company took its automotive chip subsidiary Mobileye public in 2022 while maintaining a stake, and it separated its programmable solutions group into a stand-alone business this year with plans for it to IPO within three years.
These types of moves allow Intel to refocus its resources on its core businesses after years of spreading itself too thin. An enterprise AI software platform doesn't make a ton of sense as part of Intel and would likely have proved a distraction, but by maintaining a stake in Articul8, Intel will retain some upside and will likely enjoy incremental demand for its AI-enabled chips.
While Intel has yet to experience the incredible AI-driven growth enjoyed by Nvidia, the company has more ways to win in the long run. The Articul8 spin-off adds another one to the list.
