Intel’s GPU AI Play

I was at Intel Innovation recently, and I had a chance to chat with Raja Koduri, who heads the GPU effort. Intel has announced an affordable desktop GPU option for the mid-range of graphics cards. 

Intel has one advantage in the market that Koduri is going to be betting on: It doesn’t have any high-end cards to protect. 

This means that the typical strategy of putting the most compelling features into your high-end offerings and leaving them off more value-priced efforts to protect those higher-revenue and -profit cards isn’t a strategy that Intel will execute. 

Thus, compelling capabilities that only exist in those high-end cards, strong upscaling and artificial intelligence (AI), will be available in Intel’s more attractively priced alternatives. The competitors can’t respond to this move with a like move, because they still have those high-end card margins and prices to protect. 

This turns the typical tiered pricing and product strategy on its head and allows Intel to argue you are getting much of what you’d want in an higher-priced card in a lower-priced card.  

Intel’s marketing risk

An issue for Intel will be funding the needed demand-generation marketing program that will clue users in on this value, because funding a marketing program for a value-priced part is problematic. 

If users figure this value out, it could result in a shift in market share to Intel. But that also means many users will have access to these upscaling and AI capabilities that they would not have otherwise been able to afford. 

This means that, if successful, Intel’s move into the GPU space will enable far more hardware-based AI capability at the edge. That capability will drive changes in art, science, and education.  

PC AI and productivity

Applied AI will be able to provide much better editing and creation tools. 

Intel demonstrated its text-to-art capability on stage. Using its GAUDI2 platform, Intel was able to get the AI to create a rich and accurate image by describing what they wanted the computer to create. 

Imagine being able to create your own images for papers or books, without having to worry about running down the artists or being legal issues related to images. Better, because most existing images will only approach your vision for the piece, GAUDI2 can create the perfect image to accompany your work.

AIs aren’t just useful for creating images. They are already being used to write papers from brief summaries, dramatically reducing the time it takes to issue reports. Coupled with the image creation capability, it isn’t hard to imagine a future when the need to write most papers will be passed on to our computers, which, increasingly, do both a better and faster job.  

Finally, this potentially enables the market we’ve been waiting for: intelligent digital assistants, a concept that has been presented by IBM but has not been delivered in any kind of scalable project.

This movement of GPU-based AI capability into the low end of the mid-market should provide more people with the potential to have a truly capable AI that can carry out tasks and provide answers that are more comprehensive and accurate — and even increasingly able to carry out scheduling and traveling efforts automatically.   

Disrupting the PC market

Intel’s move with its ARC graphics platform will focus compelling features, like upscaling and hardware-based AI, into more affordable PC and laptop lines. 

This will increase the potential to sell apps based on these technologies to more people at a far more affordable price. It will also make the market for AI-based apps more attractive because of the related higher sales volumes. Intel is executing a good strategy against entrenched vendors that have to protect their high-end offerings. 

Intel is releasing ARC to desktop and laptop PCs — including ARC-based laptops by Lenovo and Samsung — which will help the AI PC revolution begin once those sales get to critical mass.