The company’s overall carbon footprint for 2025 came to about 14.5 million metric tons of carbon dioxide equivalent. When comparing this to emissions by country, Google is somewhere between Ivory Coast and Panama at about 100th place.

The carbon-free energy goal

Google has been matching 100 percent of its electricity consumption with renewable energy purchases every year across its global operations, a practice that it has adopted for nine years running. In 2025, the company’s purchase agreements encompassing 12 gigawatts of “net-new clean energy” represented the largest annual total in the company’s history.

But it also acknowledged that companies can merely claim to rely on 100 percent renewable power while still using power generated by fossil fuels in local power grids. Accordingly, Google says it has renewed its emphasis on a “24/7 carbon-free energy ambition” that gets into more granular accounting by focusing on certificates representing hourly and local matches for clean energy.

To its credit, Google has become one of the world’s largest investors in clean energy technology, according to Michael Thomas, CEO of the Cleanview data platform that tracks renewable energy and data center projects. But his analysis of Google’s data center power strategy suggests that Google has pivoted to an “Everything Everywhere All at Once” strategy that encompasses everything from renewables to natural gas generation.

Google’s broader investments include efforts to commercialize “advanced nuclear, fusion energy, enhanced geothermal, long-duration energy storage, and natural gas with carbon capture and storage.” The company’s sustainability report highlighted investments of more than $3.8 billion between 2010 and 2025 that are expected to bring 7.5 gigawatts of clean energy online.

However, in an April 2026 newsletter, Thomas pointed to Google’s $40 billion investment in Texas data centers as including a campus potentially powered by a 933-megawatt natural gas power plant without carbon capture technology. The natural gas plant’s turbines could produce 4.5 million tons of carbon dioxide annually.

A Google spokesperson told Thomas that the company has not yet signed an agreement for how much power Google’s data center might draw from the natural gas plant.

This story was updated on July 2, 2026 to add more context on Google’s total carbon footprint and emissions.

https://arstechnica.com/ai/2026/07/googles-ai-buildout-drove-37-increase-in-electricity-use-in-2025/

L’AI mette sotto pressione gli obiettivi climatici di Google: consumi record di energia elettrica ed acqua nel 2025

L’intelligenza artificiale (AI) sta cambiando il mondo digitale, ma sta anche cambiando, e non in meglio, il profilo ambientale delle Big Tech. A confermarlo è l’ultimo Environmental Report di Google, che fotografa un anno in cui la corsa alla costruzione di infrastrutture per l’AI ha portato consumi di energia elettrica, utilizzo di acqua ed emissioni di gas serra ai livelli più alti mai registrati dall’azienda.

Il dato più significativo riguarda proprio l’energia. Nel 2025 la domanda di elettricità di Google è aumentata del 37% rispetto all’anno precedente, dopo un incremento già molto elevato del 27% registrato nel 2024. Se si guarda al periodo pre-boom dell’intelligenza artificiale, il confronto è ancora più eloquente: oggi il fabbisogno energetico dell’azienda è circa tre volte e mezzo superiore rispetto al 2019.

Non meno significativo è il dato sull’acqua. I consumi idrici hanno raggiunto 10,9 miliardi di galloni, con un incremento del 34% rispetto all’anno precedente e un volume più che raddoppiato rispetto al 2021. Gran parte di questa crescita è attribuita ai data center, dove l’acqua viene utilizzata per il raffreddamento delle infrastrutture informatiche, una risorsa sempre più critica soprattutto nelle aree soggette a stress idrico.

L’efficienza energetica non compensa più i consumi, aumentano le emissioni di gas serra

La crescita dell’infrastruttura necessaria per addestrare ed eseguire i modelli di intelligenza artificiale sta infatti superando i miglioramenti ottenuti sul fronte dell’efficienza dei data center. In altre parole, i sistemi consumano meno energia per unità di calcolo, ma il numero di server, acceleratori e chip installati aumenta a una velocità tale da annullare i benefici ottenuti.

Anche il bilancio climatico peggiora sensibilmente. Le emissioni di gas serra sono cresciute del 18% in un solo anno, il maggiore incremento annuale mai riportato da Google. A pesare è soprattutto la produzione dell’hardware destinato all’AI, dai processori specializzati ai server che popolano i nuovi data center. Non si tratta quindi soltanto dell’energia necessaria per far funzionare queste strutture, ma anche dell’impatto industriale legato alla loro costruzione.

L’AI manda all’aria gli impegni di sostenibilità delle Big Tech, una “buona narrazione” non basta più

Questi numeri raccontano una contraddizione sempre più difficile da ignorare. Google è da anni una delle aziende tecnologiche che investono maggiormente nelle energie rinnovabili e continua a presentarsi come protagonista della transizione energetica. Nel solo ultimo anno ha sottoscritto accordi per circa 12 gigawatt di nuova capacità da fonti pulite, un record per il gruppo, riuscendo inoltre a mantenere pressoché stabile la quota di elettricità priva di emissioni di carbonio utilizzata per alimentare le proprie attività. Le emissioni legate all’elettricità sono perfino diminuite del 3% rispetto all’anno precedente.

Tuttavia, questi risultati non bastano più a compensare la crescita impetuosa della domanda di calcolo imposta dall’intelligenza artificiale. Lo stesso report riconosce apertamente questa realtà, sottolineando che la rapida espansione del fabbisogno energetico rappresenta una sfida da gestire attivamente e che la crescita dell’AI non può diventare un pretesto per abbassare gli standard ambientali.

È proprio qui che emerge il paradosso delle Big Tech, come spiegato da Amy Harder su Axios. Per anni i rapporti di sostenibilità sono stati soprattutto strumenti attraverso i quali raccontare, “attraverso una buona narrazione”, i progressi nella riduzione delle emissioni e negli investimenti in energia pulita. Oggi, invece, gli stessi documenti ufficiali mostrano come la rivoluzione dell’intelligenza artificiale stia mettendo sotto pressione quegli obiettivi climatici che le aziende avevano indicato come prioritari.

Obiettivi di sostenibilità al ribasso, mentre cresce l’impatto dei data center sui territori: il prezzo da pagare per l’AI?

L’impressione è che il parametro di riferimento sia cambiato. Non si parla più di ridurre le emissioni in valore assoluto, ma di limitarne la crescita rispetto a quella che sarebbe stata senza gli investimenti in fonti rinnovabili. Una differenza sostanziale che modifica il significato stesso degli impegni ambientali.

Il tema assume un peso ancora maggiore perché la costruzione di nuovi data center è ormai al centro del dibattito pubblico e politico soprattutto negli Stati Uniti. In molti territori le comunità locali guardano con crescente attenzione a infrastrutture che richiedono enormi quantità di elettricità e di acqua, incidendo sulle reti energetiche, sulle risorse idriche e sull’uso del territorio.

Secondo un recente rapporto dell’United Nations University Institute for Water, Environment and Health (UNU-INWEH), entro il 2030 i data center mondiali potrebbero arrivare a consumare 945 terawattora (TWh) di elettricità all’anno. Per comprendere l’ordine di grandezza, si tratta di un volume quasi triplo rispetto al consumo elettrico annuo combinato di Pakistan, Bangladesh e Nigeria, tre Paesi che insieme ospitano oltre 650 milioni di persone.

Una transizione green che perde credibilità

La domanda non riguarda più soltanto i benefici economici e tecnologici dell’intelligenza artificiale, ma anche il prezzo ambientale necessario per sostenerne la diffusione. Prendiamo l’ondata di calore record che ha travolto tutta l’Europa in questo inizio estate, secondo diversi studi condotti negli Stati Uniti, i data center hanno un impatto sul microclima locale da non sottovalutare in termini di aumento delle temperature medie.

Google continua a sostenere che l’AI potrà contribuire, nel medio periodo, a ridurre le emissioni in altri settori dell’economia, dall’ottimizzazione delle reti elettriche all’efficienza industriale. Almeno per ora, però, il bilancio ambientale raccontato dai dati ufficiali va nella direzione opposta.

Paradossalmente, i report di sostenibilità delle Big Tech stanno diventando uno degli indicatori più osservati per capire se la promessa di uno sviluppo digitale sostenibile riuscirà davvero a convivere con la gigantesca espansione infrastrutturale richiesta dall’intelligenza artificiale.

Novità su Google, per aggiungere Key4Biz tra le tue fonti preferite, clicca qui

Leggi le altre notizie sull’home page di Key4biz

https://www.key4biz.it/google-aumenta-emissioni-di-co2-18-consumi-energetici-37-ed-idrici-34-per-colpa-dellai/578369/

Google is so famous for killing products that there’s a whole virtual graveyard you can explore. Google’s latest shutdown now has a headstone of its own. Effective today, Google has discontinued the Tenor API, which you may not be familiar with by name. You’ve probably used it, though. Tenor is a database of searchable GIFs, which used to serve animated images to sites like X/Twitter, Discord, and more. Now, it only serves Google—maybe the headstone is a bit premature.

Like many Google products, Tenor started as an independent company. Google came along and bought Tenor in 2018, and it continued running it largely unchanged in the intervening years. Tenor was integrated into Google products like Gboard and Google Messages, but the API also gave other platforms a way to help users find, share, and save GIFs. It’s similar to services like Giphy and Klipy.

In January, Google announced it was going to start winding down that API access. It stopped accepting new integrations at that time, and the end date has now arrived: As of June 30, the Tenor API is no more. Google, a company with nearly 200,000 employees and more than $130 billion in 2025 profit, says it decided to stop supporting the image API so it could better focus its resources. The real problem was probably that Tenor was free, and Google didn’t see a way it could make money from a GIF API.

https://arstechnica.com/gadgets/2026/06/google-kills-tenor-gif-api-forcing-changes-at-x-discord-and-more/

There are plenty of AI image-generation models these days, but the ones capable of quality outputs tend to be slow and expensive. Google DeepMind says its new image model, known as Nano Banana 2 Lite, offers the best balance of quality and speed. It’s available today across the Google ecosystem, creating images in a fraction of the time it takes Google’s beefier models.

The new model is part of the Gemini 3.1 family—it’s technically called Gemini 3.1 Flash Lite Image. On one hand, Google says this model is ideal for exploring ideas and “rapid-fire” prototyping, applications in which quality can take a backseat. However, the company has also provided some examples aimed at showing how close Nano Banana 2 Lite can get to the quality of its other image models.

A comparison of Nano Banana 2 Lite with the non-Lite version.

Credit: Google

In addition to the examples, Google also has Elo scores from Arena.ai ready to go, showing that users rate Nano Banana 2 Lite outputs almost as highly as the non-Lite versions. However, vibemarking doesn’t always focus on the details that can make AI images look silly upon closer inspection. Google notes that Nano Banana 2 Lite tends to have more trouble with text, particularly if it’s very small, and infographics are more likely to include incorrect data. Characters and people may also show poor consistency across iterations.

Credit: Google

But it’s fast. You can go from text to an image in about 4 seconds in the default low-thinking mode. Generating the same images in the standard Nano Banana takes about 20 seconds. The speed and efficiency mean developers accessing Nano Banana 2 Lite via the API will pay a lot less. Google says it averages $0.034 per 1K image. The API rates are $0.25 for 1M input tokens and $1.50 for 1M output tokens. That’s half the rate for Nano Banana 2. For Nano Banana Pro, the input tokens are only a little more at $2 for 1M, but the output pricing is $12 (eight times higher).

https://arstechnica.com/ai/2026/06/googles-new-nano-banana-2-lite-image-model-is-its-fastest-and-cheapest-yet/

Google Finance is not a new product—it has been around for 20 years, long enough that it initially relied on Flash to display charts and graphs. The website has gotten a few major updates over the years, but it has never had a mobile app until now. Google has released the first standalone app for Google Finance, which is currently exclusive to Android, with iOS planned for later this year.

The app is available globally in the Play Store, but that’s not the only update to Google’s financial tracker. The AI-powered makeover for the Finance website is also leaving beta, making Google’s chatbot a core part of the experience. Naturally, the mobile app includes a heaping helping of generative AI that aims to make sense of irrational financial markets.

If you’ve checked out the new Finance web experience, you’ll see a lot of familiar features in the app. You can create watchlists, monitor real-time market data, and keep up with financial news in one place. While perusing graphs of stock performance, Finance will use AI to generate “key moments” that can explain why the numbers changed. This feature initially launched in the Finance web interface in May.

The mobile app also gets Google’s new AI research tool, accessible via the “Ask” button floating at the bottom of the UI. This allows users to converse with Google’s money-tuned bot about stocks. The bottom bar also includes a History section where you can easily access your past chats.

https://arstechnica.com/google/2026/06/google-finance-finally-gets-a-mobile-app-as-ai-powered-overhaul-leaves-beta/

• Google (Google Play)
• Honor (HONOR App Market)
• OPlus (OPPO App Market)
• Samsung (Galaxy Store)
• Transsion (Palm Store)
• vivo (V-Appstore)
• Xiaomi (GetApps)

Developers will also have access to new APIs to make registering as an external developer less arduous. In the coming months, Google will release an Android Developer ID Status API that will check if a package name is already registered with Google. The Android Developer Console API will let you register and manage your app package names without leaving your development environment, too.

The countdown begins

The next step toward verifying apps will come this month as Google deploys a new system service on most certified devices. The package (com.google.android.verifier) will appear on phones and tablets running Android 8 or higher, allowing Google to block the installation of unverified apps. It will remain dormant until verification is activated in your specific region.

Credit: Google

In July, Google plans to roll out the new developer APIs and begin testing for “limited distribution” accounts. This is Google’s solution for hobbyists who want to make their own apps and share them with a small group. Limited accounts won’t require a fee or government ID verification, but you can install these apps on up to 20 devices.

In August, the advanced flow will become available globally ahead of verification becoming mandatory in the first markets. As detailed a few months ago, the advanced flow will allow users to bypass verification, but the process isn’t easy. You’ll have to navigate to a buried menu, confirm you understand the risks multiple times, and wait a whole day before completing the process.

And that brings us to September, when Android devices in Brazil, Indonesia, Singapore, and Thailand will begin checking verification status before installing apps. However, things get murky after that. Google will undoubtedly monitor how verification works as millions of users are suddenly limited to verified apps, which could affect how it moves forward. Google says it intends to expand developer verification in 2027, eventually making it a global device policy.

https://arstechnica.com/gadgets/2026/06/google-shares-updated-timeline-for-rolling-out-android-developer-verification/

Google permetterà ai siti web di escludersi dai risultati generati dall’intelligenza artificiale in Search. La novità riguarda AI Overviews e AI Mode, le funzioni che sintetizzano contenuti e rispondono alle query degli utenti usando sistemi generativi.

La società lo ha annunciato in un post sul suo blog ieri. L’avvio dei test, ha spiegato, consentirà ai proprietari dei siti di decidere se le proprie pagine potranno apparire nelle risposte AI e contribuire a fondarle. Il test partirà inizialmente con un piccolo gruppo di proprietari di domini nel Regno Unito, prima di un’estensione globale.

“Siti che scelgono l’opt-out non riceveranno traffico o impression dalle nostre funzionalità di AI generativa”, ha spiegato Google. La società ha precisato anche che questa scelta “non sarà usata come segnale di ranking per i risultati di ricerca al di fuori di queste funzionalità generative di Google Search”.

Il nodo del traffico agli editori

La novità di Google annunciata ieri arriva dopo le pressioni e le preoccupazioni dell’editoria internazionale. Secondo un’inchiesta del Wall Street Journal, le funzionalità AI di Google, come AI Overviews e gli strumenti conversazionali basati sull’intelligenza artificiale, stanno incidendo in modo significativo sul traffico dei siti di notizie.

Secondo i dati Similarweb citati dal Journal, per il New York Times la quota di traffico da ricerca organica verso i siti desktop e mobile è scesa al 36,5% nell’aprile 2025, rispetto al 44% di tre anni prima.

Google sostiene invece che AI Overviews abbia aumentato il traffico complessivo sulla piattaforma, ma per molti editori il punto resta la distribuzione di quel traffico e il rischio che i contenuti giornalistici vengano usati per generare risposte senza produrre visite equivalenti ai siti originali.

Pressione dei regolatori britannici

Anche le autorità britanniche hanno fatto pressioni a Google. La Competition and Markets Authority, CMA, ha imposto all’azienda una nuova regola legata al suo potere di mercato, dopo averla qualificata come società con “strategic market status”.

Secondo la CMA, il nuovo controllo darà agli editori, incluse le organizzazioni giornalistiche, una posizione più forte nella negoziazione degli accordi sui contenuti con Google.

Già a gennaio il governo britannico aveva annunciato l’intenzione di obbligare Google a introdurre un meccanismo di opt-out per offrire “un accordo più equo” agli editori, in particolare alle testate giornalistiche. A marzo Google aveva risposto spiegando di voler sviluppare aggiornamenti per consentire ai siti di escludersi in modo specifico dalle funzioni generative nella ricerca.

In Italia si è mossa anche l’Agcom

In Italia il tema è già arrivato sul tavolo dell’Agcom. Dopo una segnalazione della FIEG, l’Autorità ha deciso di trasmettere alla Commissione europea una richiesta di valutazione sui servizi AI Overviews e AI Mode di Google Ireland, alla luce degli articoli 27, 34 e 35 del Digital Services Act.

Secondo gli editori, l’introduzione dei riepiloghi AI rischia di ridurre visibilità e reperibilità dei contenuti giornalistici, con effetti sulla sostenibilità economica delle testate, soprattutto più piccole e indipendenti, e sul pluralismo dell’informazione. Agcom ha inoltre deciso di istituire un tavolo permanente tra Google, piattaforme interessate ed editori sui temi di copyright, intelligenza artificiale ed equo compenso.

Novità su Google, per aggiungere Key4Biz tra le tue fonti preferite, clicca qui

Leggi le altre notizie sull’home page di Key4biz

https://www.key4biz.it/google-consentira-ai-siti-web-di-escludersi-dai-risultati-di-ricerca-ai-overviews-e-ai-mode/574148/

UK regulators today ordered Google to put clearer attributions and links to publishers’ content in its AI-generated search features. The UK’s Competition and Markets Authority (CMA) also said Google must give publishers a way to opt out of AI features in search.

“In a world first, publishers will now have effective tools to prevent their content being used to power AI features in search, such as AI Overviews,” the CMA said today. “This will put publishers, like news organizations, in a stronger position to negotiate content deals with Google. To boost consumer trust, Google is also now required to make sure that publisher content is properly attributed, using clear links, in AI‑generated search results.”

The CMA ruled that Google may not penalize publishers for opting out of AI, meaning that Google can’t downrank opted-out publishers in general search results. The CMA said Google will have nine months to comply with all requirements but that the agency “expects important parts of the controls to become available to publishers well before that deadline. Google will also be required to submit and publish compliance reports, supported by key data and metrics, explaining changes it has made and how it has complied.”

Google’s AI Overviews tend to give confident-sounding responses to search queries, but the links to sources in the AI Overviews may or may not support those confident responses. Clearer attribution and links could make it easier for searchers to determine the accuracy of AI Overview summaries.

The CMA applied the rules to Google after determining that it has “strategic market status” in general search services, and has ongoing investigations into Apple and Microsoft. Google today said it will comply with the CMA decision.

https://arstechnica.com/tech-policy/2026/06/google-ordered-to-put-clearer-links-in-ai-search-and-let-uk-publishers-opt-out/

Gemma 4 12B is almost as capable as the version with 26 billion parameters.

Credit: Google

Google says the new model is capable of complex multistep reasoning and agentic workflows that previously required the larger Gemma variants. Despite the smaller parameter count, Gemma 4 12B comes with the newly devised Multi-Token Prediction (MTP) drafters, which take advantage of unused processing cycles to calculate possible future tokens. The result is greater speed and efficiency. Google has released optional MTP versions of the other Gemma 4 models, but this is the first one to have MTP out of the box.

Gemma 4 12B is also more efficient thanks to a new approach to multimodality. The Gemma 4 family is natively multimodal, accepting text, audio, or images as inputs. Most gen AI models—including the other Gemma 4 variants—use dedicated encoders to process non-text inputs and pass that data to the LLM. This works well enough, but it increases latency and memory usage.

With the new mid-weight model, Google has implemented a streamlined embedding module for vision, featuring single-matrix multiplication and positional embedding, which allows the data to pass to the LLM with proper spatial awareness. This eliminates the need for a bulky middleman encoder. For audio, there’s no encoding at all. The developers worked out a method of projecting the raw audio signal into the same vectors used for text tokens.

If you want to check out the new Gemma 4 model, it’s accessible without a download via tools like LM Studio, Google AI Edge Gallery, and more. But the whole idea with Gemma 4 12B is that you can run it locally and on your own terms. If you’ve got the RAM, the model weights are available for download immediately on Kaggle and Hugging Face. It’s just shy of 18GB.

https://arstechnica.com/google/2026/06/googles-new-gemma-4-open-ai-model-is-sized-for-your-laptop/

We’re expecting Android 17 to begin rolling out later this month, but first, Google has a batch of updates for the wider Android device ecosystem. As usual, some of the new features are limited to specific devices, and others require using Google’s apps. But if you don’t mind the latter, you can get automated protection from the growing threat of deepfake phone scams.

According to Google, “impersonation fraud” is one of the most common types of financial scams. The FTC tracked almost $3 billion in losses from such scams during 2024, and the improvements in AI voice cloning tools more recently are making the schemes easier to pull off. The voice models are becoming so capable that it can be difficult to identify a fake caller even when an AI is imitating someone you talk to every day.

Google’s solution is an expansion of the system it debuted last month for verified financial calls. Now, a similar feature will work with anyone in your contacts. Many of the most effective deepfake scams involve spoofing a contact’s number, which makes the call look more legitimate when your phone lights up. Victims of these scams are then greeted by an accurate re-creation of the person’s voice spinning a yarn that involves an urgent need for cash.

Google’s scam call detection feature will be available on all phones running Android 12 or higher, but it does require you to have three Google apps installed: Phone by Google, Contacts, and Google Messages. Depending on your device, you may already have these. They’re the preloaded options on Pixel and Motorola phones, and Samsung has now switched over fully to Google Messages. Google claims that Phone by Google is the most widely used dialer, but that doesn’t seem right—Samsung has its own phone app, and it’s the largest Android OEM by far.

https://arstechnica.com/gadgets/2026/06/google-announces-deepfake-call-detection-for-android-new-airdrop-device-support/