It’s no secret that AI is a hog, consuming energy and water like no digital technology before it. Now we know just how much Big Tech’s pursuit of AI is costing the environment.

Both Google and Amazon released their sustainability reports this week, and the numbers aren’t pretty. Each company has pledged to zero out its carbon emissions in the coming years, but AI has made those goals a lot harder to hit. Google’s total carbon emissions are up 25% since last year, Amazon’s are up 16%.

A close reading of the reports suggests that both Amazon and Google will have to make some serious, and potentially costly, adjustments to their businesses if they’re going to achieve their net-zero targets.

Neither company comes out and blames AI directly for the rising emissions, but there’s plenty of indirect evidence.

AI at the center of it all

Both Amazon and Google acknowledge their energy use has increased significantly in the last year as use of AI has risen. Both talk about carbon intensity — essentially, how much pollution a company generates for every dollar of revenue it brings in — a metric China has used over the last several years when negotiating climate treaties even as its emissions were skyrocketing. And both devote several pages touting how AI can benefit the environment, a case of “protesting too much,” to borrow some Shakespeare.

The picture gets clearer the deeper you dig into the data. Both companies are actually doing OK when it comes to carbon pollution from energy purchases. Years of buying renewable power have helped keep a lid on things, though that may change in the near future as tech companies, including Google, have begun to invest heavily in natural gas power plants to keep pace with AI’s power demands.

Rather, most of Amazon’s and Google’s growing carbon footprint comes from so-called Scope 3 emissions — a catch-all category covering pollution a company doesn’t directly control, like the goods and services it buys or the products it sells. For companies like Amazon and Google, Scope 3 includes things like GPU purchases and the use of a company’s products, like phones and tablets.

Google lumps together two categories of Scope 3 emissions — capital goods and use of sold products —though it admits the latter is small enough to not be material. (Most of Google’s hardware products are small devices that don’t consume a lot of electricity.) That likely leaves data centers as the main driver. Last year, Google’s Scope 3 emissions increased by 2.1 million metric tons, which means they’re now double what they were in 2019, the year Google uses as its baseline when assessing its performance.

Amazon’s rising Scope 3 emissions mostly come from capital goods and fuel and energy. The former can include data centers and warehouses, which can help explain why Amazon’s Scope 3 emissions spiked higher than Google’s. Still, a good chunk is probably data centers. “To meet strong customer demand, in 2025 we added more data center capacity globally than any other company, including more than 1.2 gigawatt (GW) in Q4 alone,” Amazon wrote in the report.

Hitting a wall

That kind of spending helps explain why decarbonization is suddenly getting so much harder. For years, the biggest contributor to their carbon footprints was energy for offices and more modestly sized data centers. That could easily be canceled out buying renewable power. 

AI has upended that approach. While tech companies could still use renewables plus batteries to power their data centers, they’re starting to fall back on fossil fuels. It’s a trend that will make their net zero pledges that much harder to deliver, but it’s not irreversible.

The more pernicious emissions come from the construction and outfitting of data centers themselves. The steel and cement industries are both heavy polluters, and while startups are working on low-to-zero carbon approaches, they’re still not ready to deliver at the scale that tech companies need. 

Then there are the GPUs and memory chips powering the AI boom. Semiconductor manufacturing uses lots of energy, and many of the world’s leading-edge chip factories are located in Asia, where the electrical grids remain dominated by fossil fuels. Making matters worse, many of the chemicals used in those factories are also potent greenhouse gases, capable of warming the atmosphere thousands of times more than an equivalent amount of CO2. The bingeing on chips has probably inflated both Amazon’s and Google’s carbon footprints.

None of these problems are intractable, though Amazon, Google, and their peers have their work cut out for them. To deliver on their net-zero pledges, they’ll need to ramp up their renewable energy purchases, invest heavily in advanced steel and cement manufacturing, and buy many millions of tons of carbon removal credits. It’s still possible, but their embrace of AI hasn’t made it any easier.

Back in April, Reuters reported that Anthropic was toying with the idea of producing its own AI chips as a means of responding to chip shortages. Now, it would appear that the company is getting serious about this idea.

On Thursday, The Information reported that Anthropic was in contact with Samsung to explore a collaboration around the pending chip. However, Anthropic hasn’t yet decided what the chip will be used for, how it will fit into the server, or how powerful it will be, according to the report.

When reached for comment, Anthropic told TechCrunch that a diversified hardware stack that includes chips from Google, Amazon, and Nvidia will continue to be pivotal to its compute strategy. On the topic of a potential Samsung partnership, the company said it had nothing further to add.

A number of AI companies have sought to develop custom chips — both as a way to create unique hardware for specific compute tasks and to gain a certain amount of independence from Nvidia, which continues to be the undisputed leader of the chip industry.

Anthropic’s announcement may also be a response to one made last week by its key competitor, OpenAI, which has teamed up with Broadcom to announce its own custom built inference processor, dubbed “Jalapeño.” OpenAI says that the chip is more efficient, demonstrating better performance-per-watt, than other competitor chips. Amazon and Google both offer custom-built TPUs as part of their cloud offering.

Samsung is already embedded in the AI industry, and acts as a major partner of Nvidia, producing chips that the company needs to train or run its AI models. In turn, Samsung uses Nvidia’s software to manufacture its chips. The duo are working on an AI chip factory in South Korea. Samsung has also discussed partnering with Google on its chip-making efforts.

Meta is getting into gaming with the launch of a new app called Pocket, which allows people to generate small, interactive apps and games using AI prompts. The software, a result of Meta’s acquisition of the team at the vibe-coded gaming platform Gizmo earlier this year, describes itself as “a creative platform for making and sharing gizmos,” which is what the interactive experiences are called. It also offers a scrollable feed where you can play with gizmos others have made.

Based on the app’s screenshots in Google Play, there are many similarities to Gizmo’s original app, which is still listed. Like Pocket, Gizmo also offers a way to use written AI prompts to build small, interactive experiences, and it includes a discovery feed.

Alessandro Paluzzi, a reverse engineer and regular spotter of new apps and features, first noticed the app’s launch this morning and published a Play Store screenshot of the app on X. According to data from app intelligence provider Appfigures, however, Pocket was first launched on June 29, 2026 on the App Store and Google Play. (Because of its newness, the firm can’t tell if it’s yet to see any downloads.)

Other outlets, including Business Insider and Investing.com, have also reported on Paluzzi’s discovery. Meta has not yet responded to a request for comment.

Pocket is another example of Meta’s push to make AI creation tools more mainstream, extending its earlier efforts, which included AI-generated images created via its Meta AI app, and AI videos created with its app called Vibes. It has also added AI features across its social platforms and into its video-editing app for creators, Edits.

Given that Meta has not officially announced Pocket’s debut, it’s likely that Pocket is still in its initial experimentation phase.

Its counterpart Gizmo, however, had generated 635K lifetime installs across both iOS and Google Play, according to Appfigures, which noted it had a 98% positive sentiment.