72627d30 056f 11ee b1eb 1585a3a8dc28

With Apple entering the fight, the AI chip wars have gone nuclear

Posted by


The artificial intelligence (AI) chip wars have officially kicked into high gear. On Tuesday, Apple (AAPL) unveiled its latest iPad Pro lineup, complete with the company’s all-new M4 chip. The processor, Apple says, features a 50% faster CPU than the prior generation Pro’s M2 chip, as well as a GPU with four times the performance of its predecessor.

But Apple focused most on the M4’s neural engine, the portion of the chip dedicated to running AI applications. According to the company, the neural engine in the M4 is the company’s fastest yet, and it’s more powerful than any neural processing unit available on today’s AI PCs.

That might all seem like Apple puffing out its chest at the competition, and it is to a degree. But it also shows we’re entering a new era of PC marketing focused on the promise of generative AI applications running directly on our machines, rather than in the cloud.

Microsoft (MSFT) is already pushing the category of AI PCs along with Intel (INTC), AMD (AMD), Qualcomm (QCOM), and Nvidia (NVDA), and with Apple entering the fray, the battle over which company is the new AI champ has officially begun.

The AI PC, or in the case of Apple AI iPad or AI Mac, is meant to be able to run smaller versions of the kinds of large language models that power apps like ChatGPT, Microsoft’s Copilot, and Google’s Gemini. In fact, Google (GOOG, GOOGL) already runs a version of its Gemini model called Gemini Nano on its Pixel smartphones.

Apple's M4 chip brings with it increased AI capabilities, which could tip what's to come at WWDC next month. (Image: Apple)Apple's M4 chip brings with it increased AI capabilities, which could tip what's to come at WWDC next month. (Image: Apple)

Apple’s M4 chip brings with it increased AI capabilities, which could tip what’s to come at WWDC next month. (Image: Apple) (Apple)

AI PCs are, generally, PCs that include neural processing units (NPUs), or in Apple’s case, a neural engine. These are essentially portions of a chip dedicated to powering AI applications. Intel’s Core Ultra chips, which it debuted in December, feature their own NPUs, while AMD has been showing off chips with its XDNA AI engine since May 2023. Qualcomm, for its part, has also joined the fight with its Qualcomm Snapdragon X Elite and X Plus chips.

All of these chips should end up in Windows PCs on store shelves by the end of the year. Microsoft is already banking on AI PCs being a runaway success, going so far as to introduce a Windows keyboard button dedicated to its AI-powered Copilot software. It’s the first new keyboard button in nearly 30 years.

Apple, on the other hand, has been building its neural engine into its chips since it launched its A11 processor in the iPhone X in 2017. The company has regularly touted its neural engine’s capabilities throughout the years, but the debut of the M4 chip marked the first time Apple started throwing out key performance metrics.

During its press event unveiling the M4, Apple was sure to point out that it not only has been outfitting its chips with neural processing engines for years but that the M4 is capable of running 38 trillion operations per second (TOPS). TOPS is more or less a measurement of how well a chip can handle AI tasks.

Qualcomm, for its part, says that its Snapdragon X Elite and X Plus chips offer 45 TOPS of performance. And according to CNET, Intel and AMD’s chips top out at 34 and 38 TOPS, respectively.

Qualcomm on Wednesday announced its latest chips designed to run on Windows laptops, taking aim at Intel and Nvidia. (Image: Qualcomm)Qualcomm on Wednesday announced its latest chips designed to run on Windows laptops, taking aim at Intel and Nvidia. (Image: Qualcomm)

Qualcomm on Wednesday announced its latest chips designed to run on Windows laptops, taking aim at Intel and Nvidia. (Image: Qualcomm) (Qualcomm)

You can’t mention AI performance without talking about Nvidia, but it’s almost unfair to compare its chips to the competition. That’s because Nvidia’s graphics chips are dedicated to graphics processing and can be programmed to run AI apps. But because they’re not multipurpose chips like those from Apple, Intel, AMD, and Qualcomm, they’re far more capable. In fact, Nvidia claims its laptop chips offer 20x to 60x the performance of competing neural processing units.

The point is, there’s a new stat for these tech companies to battle over, and you can be sure they’ll do just that. The fact that Apple had already come out swinging with its own measurement is proof enough of that.

What does that mean for the average consumer? Not all that much just yet. Generative AI apps running on PCs and Macs and iPads are still few and far between. So far, there are programs that help you manipulate photos and videos or track your eye movement in video chat apps.

But there’s still no killer app that would get me to go running out the door to pick up an AI PC. That could all change in the coming weeks, though. That’s because Microsoft and Apple are set to kick off their annual developers conferences, where they’re expected to debut a slew of AI apps and services, and at least some of those should run on their respective devices.

How much they manage to entice consumers to pony up for AI PCs and the like, however, will depend on whether they’re truly useful or not.

Subscribe to the Yahoo Finance Tech newsletter.Subscribe to the Yahoo Finance Tech newsletter.

Subscribe to the Yahoo Finance Tech newsletter. (Yahoo Finance)

Email Daniel Howley at dhowley@yahoofinance.com. Follow him on Twitter at @DanielHowley.

Click here for the latest technology news that will impact the stock market.

Read the latest financial and business news from Yahoo Finance





Source link

Leave a Reply

Your email address will not be published. Required fields are marked *