The Cupertino, Calif.-based tech giant claims that the new M4 system-on-chip inside the forthcoming 11- and 13-inch iPad Pro tablets has more powerful AI processing capabilities than processors powering any Windows-based AI PC on the market today.

Apple is using the reveal of its new, M4-powered iPad Pro tablets to claim it’s far ahead of Windows-based AI PC vendors thanks to a new version of its Neural Engine it says is “more powerful than any neural processing unit in any AI PC today.”

The Cupertino, Calif.-based tech giant on Tuesday revealed the next generation of iPad Pro, which will feature a smaller, 11-inch version in addition to the standard 13-inch model. The company also revealed a new iPad Air, which will include a new 13-inch model on top of the standard 11-inch size and use the company’s M2 chip that debuted in 2022.

[Related: AMD Says Ryzen Pro CPUs Enable Better AI PCs For Businesses]

Apple said the new iPad Pros will be its first devices to use the company’s new M4 system-on-chip, the successor to the M3 that debuted in Mac computers last fall. For the first time in iPad’s history, the premium tablet designs are also getting OLED screens, thanks to a new display technology that combines two OLED panels to increase brightness and color vibrancy.

The new iPad Pro models are supported by the all-new Apple Pencil Pro and a new thinner, lighter version of the Magic Keyboard.

Apple CEO Tim Cook called the reveal of the new iPad Pro and iPad Air designs the “biggest day for iPad since its introduction” in a video detailing the devices.

“iPad enables people to take their ideas, their work and their creativity to new heights. We can’t wait to see what users do with these incredible new iPads,” he said.

With availability expected to begin on May 15, the 11-inch iPad Pro will start at $999 while the 13-inch model will start at $1,299. The 11-inch iPad Air, on the other hand, will start at $599 while the 13-inch version will start at $799.

Apple Explains Why It’s Using The M4 For New iPad Pros

Tim Millet, vice president of platform architecture at Apple, said the company decided to use M4 and not the M3 to power the new iPad Pros because the new chip was “essential to deliver incredibly performance in the exceptionally thin and light design” of the two tablets.

The 11-inch and 13-inch iPad Pros are Apple’s thinnest devices yet, measuring at 5.3 millimeters and 5.1 millimeters respectively, according to John Ternus, senior vice president of hardware engineering. This makes them “even thinner than the iPad Nano,” he said.

Millet said the M4 brought another major benefit for the new iPad Pros: a new display engine that enables their all-new, OLED-based Ultra Retina XDR displays, which use the company’s new Tandem OLED technology to deliver 1,000 nits of sustained brightness and 1,600 peak nits.

The M4 contains a “powerful new CPU,” which comes with four performance cores and six efficiency cores, according to Millet, and all of the cores are equipped with next-generation machine learning accelerators. With everything combined, the CPU delivers up to 50 percent faster performance than the CPU in the M2-based iPad Pro, he added.

The M4 also comes with a 10-core GPU that is based on the next-generation GPU architecture that debuted in the M3, bringing features like hardware-accelerated mesh shading and ray tracing to iPad “for the first time” to enable state-of-the-art graphics, Millet said.

The executive said Apple made these improvements with the M4 while “maintaining our industry-leading performance-per-watt,” which allows the chip to “deliver the same performance as [the] M2 using just half the power.”

He added that the M4 can also deliver the same level of performance of a new x86 chip in a new thin-and-light PC—revealed to be Intel’s Core Ultra 7 155H chip in the ASUS Zenbook 14 OLED laptop in a press release—using “just a quarter of the power.”

M4 Makes iPad Pros Faster Than Today’s AI PCs, Apple Claims

Another key component of the M4 is the Neural Engine, which is Apple’s version of the neural processing unit (NPU) that accelerates AI workloads.

Ternet made a point to say that Apple’s custom processor designs have used NPUs far longer than chip vendors: “Now while the chip industry is just starting to add NPUs to some of their processors, we’ve been including our industry-leading Neural Engine in our chip for years,” he said.

While the Neural Engine debuted in iPhones in 2017 with Apple’s A11 Bionic chip and made its way into Macs with the M1 in 2020, AMD’s processors didn’t get an NPU until 2023 while Intel’s first processors with an NPU arrived later that year.

The outlier is Qualcomm, which debuted an early version of an NPU in 2015 for smartphones and a newer version for PCs in 2018. While the company’s initial efforts didn’t gain much traction, it’s making a much bigger bet on the product category with the upcoming Snapdragon X processors, which, like Apple’s M-series chips, are custom Arm designs that include an NPU.

While chip companies are making new or rekindled efforts to accelerate AI workloads in PCs with NPU, Ternet said Apple’s M4 chip features its “most powerful Neural Engine ever.”

He added that the Neural Engine is capable of an “astounding” 38 trillion operations per second, or 38 tera operations per second (TOPS), which is becoming a common way of measuring AI performance in PCs.

By contrast, Intel’s Core Ultra lineup is capable of 11 TOPS with its NPU and 35 system TOPS across the CPU, GPU and NPU while AMD’s Ryzen 8040 series can hit 16 NPU TOPS and 39 system TOPS.

“In fact, with this level of performance, the Neural Engine in M4 is more powerful than any neural processing unit in any AI PC today,” Ternet said.

However, Intel plans to boost NPU TOPs to 45 and system TOPS to 100 or more with its next-generation Lunar Lake processors set to come out later this year. AMD plans to more than triple the NPU performance with its next-generation Strix Point processors coming in the second half of this year compared to its Ryzen 7040 series, whose NPU can hit 10 TOPS. Qualcomm, on the other hand, said the NPU in its upcoming Snapdragon X processors will deliver up to 45 TOPS, which will enable the chips to hit a system TOPS of 75.

While Apple didn’t say how many system TOPS the M4 can deliver, Ternet said that the M4 is an “outrageously powerful chip for AI” when combining the processing capabilities of its CPU, GPU and Neural Engine as well as the chip’s improved memory bandwidth.

For example, the M4’s horsepower can let a user “easily isolate a subject from its background in 4K video with just a tap in Final Cut Pro,” according to Ternet.

Apple Fans Will Have To Wait Longer For Promised GenAI Features

While Apple talked up the AI capabilities of its new M4 system-on-chip, the company did not share any details about generative AI features coming to its devices.

Just last week, Cook promised that the company would have an announcement about generative AI capabilities to share in the “weeks ahead.”

By the end of Apple’s iPad presentation on Tuesday, it was clear the video was not the planned venue for any announcements related to generative AI.

However, Cook said at the end of the presentation that Apple plans to discuss the “future of our platforms and share exciting details about what’s to come” at its Worldwide Developers Conference, which kicks off next month on June 10.

In contrast to vendors in the Windows PC ecosystem, Apple had remained relatively quiet about the AI capabilities of its devices for a broad audience until the company in March called its newly released MacBook Air the “best consumer laptop for AI” and said every Mac with an M-series chip is a “great platform for AI,” according to a previous CRN analysis.

“We continue to feel very bullish about our opportunity in generative AI we are making significant investments, and we’re looking forward to sharing some very exciting things with our customers soon,” Cook said in Apple’s earnings call last week.