Wednesday, June 12, 2024

Intel and the AI Race

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

I recently attended a briefing on artificial intelligence (AI) hosted by Kavitha Prasad, the VP and GM of Intel data center, AI, and cloud execution and strategy. 

Intel was one of the first component companies to begin work on AI and had some impressive early examples of applying the technology to things like drone swarms and headlights that could penetrate snow and rain. Intel had advanced researchers, like Brian David Johnson, who used science fiction prototyping to better anticipate our AI future. But Intel took its foot off the gas.

Intel isn’t out of the race, however, as it recently won the AI autonomous vehicle DARPA challenge, along with Microsoft, again. They were part of a team that won a prior challenge years ago as well. Over time, Intel has proven its advantages, though it can improve on training. 

Given the iffy nature of AI now, Intel has time, if it can outperform its competitors. It is going to be an expensive race. 

See more: Top Performing Artificial Intelligence (AI) Companies

Why Intel has time

As Prasad shared in her presentation, AI is still in its infancy, with around 70% of AI implementations failing to meet their stated objectives. This is similar to the rollout of client/server computing, which also had a ton of hype initially but after a decade, was still failing much of the time. It takes a while to get the perfect balance of skills, technology, and practical knowledge for a different and complex technology to truly mainstream. 

Yes, we have had several successful AI trials and deployments largely surrounding mature enterprise class AI, but even those suffered through long periods where they underperformed financially, because the cost of training the system was so high. In addition, problems with data access have proven a huge problem with spinning up a truly helpful AI solution in health care, where it is most needed. Federated data and synthetic alternatives are beginning to make a dent in both data access and training costs but with the potential of increased bias, which then must be mitigated to assure an effective and reliable result.  

However, this window isn’t unlimited. It’s expected that by the second half of the decade, the next generation of AIs, broad AIs, are expected to mature, promising both an increase in flexibility and a reduction in related costs. And with employee claims of sentient AI, which seems doubtful, this market isn’t slowing. It’s speeding up, a lot.  

So Intel has time but not a lot of it. Intel Labs is an impressive organization, and Pat Gelsinger is arguably one its best CEOs, so it has a decent shot at success.  

A wide-open market

The AI market is anything but mature which makes it anyone’s game for now, including Intel.

However, Intel’s competitors are increasing, and the company is having to fight to build out more manufacturing capacity without adequate support from the U.S. government. Gelsinger has an impressive number of critical things that need doing on his plate, but if there is anyone up to that task, it is Gelsinger.  

Given the recent DARPA win, while Intel is far from a shoo-in, it’s motivated to get there. And this competition is likely to result in some unconventional approaches that could break this market wide open. Regardless of who wins, if we get AI right, it will make a massive difference in the world. 

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles