Apple Just Surprised Everyone With the M4 Chip

Apple’s iPad announcements are usually quick, low key affairs: Update the screen, put the latest MacBook chip in the iPad Pro, throw in some skits for razzle dazzle, and call it a day. But today’s iPad event surprised everyone by revealing the next Apple Silicon chip: the M4.

In an unprecedented move, the iPad Pro is actually leaping ahead of the MacBook by getting the first crack at Apple’s newest in-house chip. While it’s not a straight ahead replacement for pro-level silicon like the M3 Ultra, the M4 coming to the iPad Pro first means it’s feasible your next iPad will be more powerful than your current MacBook Air (or 14-inch MacBook Pro for that matter).

This all has major implications for Apple's place in the ongoing big tech AI war, where it has historically lagged behind.

What’s different about Apple’s M4 chip?

The Apple M4 chip is, predictably, all about AI. While M-series chips have always had a neural engine (or NPU) built-in, the M4’s neural engine is getting a massive efficiency boost. It’s still just 16 cores, but Apple claims it can now run “38 trillion operations per second,” a purported sixty times speed improvement over the company’s first neural engine. By comparison, the M3’s neural engine topped out at 18 trillion operations per second.

“The neural engine in M4 is more powerful than any neural processing unit in any AI PC today,” said Tim Millet, VP of Apple’s Platform Architecture division.

That’s a big claim for a chip that’s debuting in a tablet, and one that may not be true for too much longer (more on that later). But the M4 is also improving in more traditional ways: In addition to its four performance cores, the M4 uses six efficiency cores, two more than on the M3. Its 10-core GPU is largely the same on paper as the M3’s, although Apple claims four times faster rendering performance than on the M2—a many fold increase over the claims it made with the M3.

Apple is also planning to continue leading the industry in power efficiency. “M4 can deliver the same performance as M2 using just half the power,” Millet claimed.

Rounding out these improvements is a new display engine, built largely to support the iPad Pro’s OLED screen. This engine will power the device’s 10Hz-120Hz dynamic refresh rate screen, plus aid in brightness and color compensation. Brightness is a typical pain point for OLED, and is something the iPad Pro is trying to fix with its new “tandem OLED” screen, which essentially stacks two OLED displays on top of each other. The display engine will also aid in keeping these screens in sync.

What does M4 mean for Apple AI?

All eyes are on Apple's forthcoming WWDC this June, where the company is finally expected to announce its AI competitor to the likes of ChatGPT and Google Gemini. The M4 chip debuting in a tablet first only lights a fire under these rumors.

With the launch of (and subsequent disappointment in) standalone AI devices like the Humane AI pin and Rabbit R1, it’s clear the market is aching for AI implementation that moves beyond novelty and actually integrates into your mobile operating system. Such an AI assistant could easily set appointments, change phone settings, send texts, and more. Google is expected to be the first to bat with such an AI at Google I/O next week, but the next month could easily see Apple’s mobile operating systems following suit. It’s unclear what the Cupertino developer has in store for iPhone, but when viewed in this light, it makes sense that M4 is coming to the iPad before the MacBook. Putting such a powerful neural engine in the iPad sets Apple's tablets up for success in what is likely to be the next big mobile OS battlefield.

Prior to WWDC, the neural engine in the M4 chip will continue to do what it always has–enable some fun magic tricks in Apple-developed programs. “It can do amazing things even faster,” Millet said. “Like easily isolate a subject from its background in 4K video with just a tap in Final Cut Pro.”

That's impressive, but I'm looking forward to finally seeing that kind of power applied to more robust ends. Until now, the M-series neural engine has come across as a bit of future-proofing, with most AI relegated to the cloud, rather than running locally. The M4 sets Apple up for the next stage of on-device AI.

When will the M4 chip come to the MacBook?

Apple’s M-series chips aren’t just for the iPad. More traditionally, they start off in MacBooks later before coming to the company’s iPads. While Apple doesn’t tend to announce new MacBooks until closer to the fall, the M4 debuting so early sets the groundwork for what the next generation of MacBooks will look like.

First, I;m calling it: Get ready for OLED MacBooks. OLED coming to MacBook has been hotly anticipated for years, as it is already a mature technology when it comes to PC laptops. With the M4 featuring a display engine built specifically to support OLED, there’s little doubt the next line of MacBooks will follow in the iPad Pro’s footsteps later this year.

OLED might even come to the MacBook Air, despite being reserved for the more expensive model of iPad, as the MacBook Pro tends to rely less on the base M-series chip and more on pro-level refreshes, which in this case would be the M4 Pro, M4 Ultra, and M4 Max. The new MacBooks will also likely integrate with whatever mobile-first AI initiatives Apple announces at WWDC. This would provide a quick way for the iPhone maker to establish a niche for itself that Google and Microsoft can’t, as it makes both full-fledged computers (sorry, Chromebooks) and smartphones.

What about AI on Windows?

Even as Apple is working to set the stage for a big AI showcase later this summer, it’s still going to have to play catchup. After Google I/O on May 14, Microsoft has told media that it will be holding a Surface AI event in Seattle on May 20. There, the company will share its “AI vision,” set to focus on Windows on Arm.

Sources “familiar with Microsoft’s plans” told The Verge in April that the company is confident its new Arm-powered Windows laptops will beat the M3 MacBook Air in CPU performance and AI tasks. Even with the M4 chip out now, that represents a significant threat for Apple—Arm is the same architecture powering Apple Silicon, and while it tends to lag behind Intel and AMD chips when it comes to power, it is usually far more efficient. If Microsoft can catch up to Apple on battery life without sacrificing much power, it will take away one of the few remaining hardware advantages MacBooks have over the much more diverse array of Windows machines, especially with M4 being limited to tablets until later this fall.

It remains to be seen just how much power we can expect from Microsoft's new Windows on ARM machines, which are reportedly powered by Qualcomm's Snapdragon X Elite.