Apple’s unique (and failing) AI strategy helps it achieve net zero, maybe?

By now, we all know that the AI bubble poses a big challenge for decarbonisation. In a rush to build gigawatt-scale AI data centres, big tech has reversed years of steady progress towards net zero, as Claire Brown recently wrote in the Climate Forward newsletter for the New York Times:

The artificial intelligence boom could pose a serious threat to tech company goals to zero out emissions by the end of the decade, according to a fresh batch of sustainability reports.

Google’s greenhouse gas emissions rose by 11 percent in 2024 from the year before. Amazon’s were up by 6 percent. Microsoft’s fell slightly but remained 10 percent higher than they were in 2021. Meta’s most recent figures have not yet been made public.

You don’t need to squint hard to notice one company, Apple, is conspicuously missing from the familiar list of big tech names here. According to Apple’s 2025 Environment Progress Report published in April, its total gross carbon footprint (scopes 1, 2, and 3) continued to decrease from 2023 to 2024.

In this AI bubble, some of Apple’s standout progress can be attributed to its unique, and very publicly failing AI strategy. Unlike its big tech competitors who rush to build bigger data centres to run larger AI models on the cloud, Apple chose to prioritise “on-device” AI capabilities. On Apple devices, most AI tasks are carried out by smaller, specialised models running on Apple’s advanced proprietary chips. Occasionally, more complex tasks will be kicked up to Apple’s Private Cloud Compute data centres, or farmed out to ChatGPT.

Naturally, with this strategy, Apple needs less cloud infrastructure than its competitors. This means less steel, less concrete, less coolants, and less electricity. Neat.

But! These smaller models running on highly efficient Apple silicon chips in millions of Apple devices still consume energy. Instead of using all the energy to do all the AI in data centres, each Apple Intelligence-enabled device uses a little bit of energy to do a little bit of AI on-device. Does this mean all these devices, collectively, will generate much more emissions because of AI?

Apple’s product environmental reports might provide some clues. iPhone 15 Pro was announced in September 2023, nine months before Apple announced its AI features at WWDC 2024. Three months later, Apple announced its “built-for-Apple-Intelligence” iPhone 16 Pro. These two models, announced one year apart, have the same design and materials, so it’s unsurprising that they also have the same total lifecycle product carbon footprint at 66 kgCO₂e for the 128GB configuration. But for 16 Pro, the product use phase (i.e., charging) contributes 17% of its total product footprint, whereas only 15% of 15 Pro’s footprint comes from product use. That’s a 13% increase from one generation to the next.

I have no reason to think the 16 Pro is somehow 13% less energy efficient than the 15 Pro. My theory is that after announcing its AI features at WWDC 2024, Apple expected customers to do more AI on the iPhone 16s, which would in turn consume more energy in the use phase.

If my theory is right, we will soon see a meaningful increase in Apple’s total scope 3 emissions, especially from the product use phase.¹

Or maybe not? As we now know, Apple still hasn’t shipped many of the AI features it promised in June 2024, and those they shipped aren’t very useful. Maybe Apple’s scope 3 product use emissions won’t increase yet because people are not doing AI on their iPhones yet?

So what does this all mean for Apple’s net zero by 2030 target? For now, it seems like Apple’s on-device AI strategy and inability to ship useful AI features are helping them avoid emission surges that have been inflicted upon others. But if Apple is able to overcome its AI challenges and people start doing more AI on-device, Apple will have a bigger net zero problem than its competitors. After all, it’s probably easier to buy renewables for a few ginormous data centres, than doing the same for hundreds of millions of users charging their iPhones in every corner of the world, at all hours of the day.